Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Improved method for evaluating the quality of phenolic resin bonds of Douglas fir Northcott, Philip Lachlan 1954

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1954_A7 N63 I6.pdf [ 8.59MB ]
Metadata
JSON: 831-1.0106855.json
JSON-LD: 831-1.0106855-ld.json
RDF/XML (Pretty): 831-1.0106855-rdf.xml
RDF/JSON: 831-1.0106855-rdf.json
Turtle: 831-1.0106855-turtle.txt
N-Triples: 831-1.0106855-rdf-ntriples.txt
Original Record: 831-1.0106855-source.json
Full Text
831-1.0106855-fulltext.txt
Citation
831-1.0106855.ris

Full Text

IMPROVED METHOD FOR EVALUATING THE QUALITY OF PHENOLIC RESIN BONDS OF DOUGLAS FIR by PHILIP IACHIAN NGRTHCOTT A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in the Department of Forestry Faculty of Applied Science We aooept this thesis as conforming to the standard required from candidates for the degree of MASTER OF APPLIED SCIENCE Member of the Department of Forestry Member of the Department of Mechanical Engineering THE UNIVERSITY OF BRITISH COLUMBIA A p r i l , 1954-- 1 -ABSTRACT The purpose and history of glue bond testing have been reviewed, .Certain deficiencies of standard test methods have been elaborated, and the need for a more aoourate test procedure stated* The objective of this par-ticular research was to develop a test that would meet such a need. Exper-imentation was limited to Douglas f i r veneers bonded with hot-press phenolic resin adhesives. The following requirements of an ideal method of estimating glue bond quality were used as a guide i n selecting new designs. The foremost requirement i s reproducibility of test results. Other essential features include a universally acceptable unit of measurement and a test speoimen which i s simple and economical to prepare. Other desirable features are that a maximum number of specimens be obtainable from a given sample, and that the method be adaptable to both research and production testing. The Glueline-Cleavage Test developed through this research meets a l l of the above requirements. The principle i s to measure the force re-quired to s p l i t or cleave a one inch square plywood specimen along the glue line by means of a "knife" or wedge. The aotion ia similar to the s p l i t t i n g of any wood with a wedge except that the knife i s particularly positioned along the glue l i n e . Specimens for this purpose may be either cross-banded or laminated. I f the data are to be directly comparable every detail of the specimen, manufacture, testing machine, and test procedure must be standard-ized. This requirement i s common to a l l methods of test. For quality oon-t r o l purposes test specimens would be out from the plywood at an angle of forty-five degrees to the grain. Every glue line of the speoimen may be tested. Special two-ply specimens have been designed for research purposes where aocuracy is of the utmost importance. In this type, the material i s - 2 -cut so that the cells interseot the veneer surfaoe at an angle of ten degrees. This small angle, plus the relative weakness of wood i n tension perpendioular to the grain, tends to concentrate the stress i n the glue li n e . This insures a glue line failure when the knife i s applied i n the correct direction. If the knife i s not applied i n the correct direction the specimen tends to s p l i t along the grain of the wood away from the glue line rather than toward i t as intended. These research specimens are pre-pared from two edge-grain veneers glued with the springwood-summerwood bands crossed. Any cross-banding angle up to ninety degrees (common for oommeroial plywood) may be used. The strengths of matched phenolic-bonded Douglas f i r glue lines were compared by five methods, four being mechanical and one relying on wood failu r e . The mechanical methods were the Blook Shear, Glueline-Cleavage, Tension Normal to the Glue Line, and Tension Shear Tests. The Per Cent Wood Failure Method employed wood failure readings from the Ten-sion Shear specimens as estimates of bond quality. The Glueline-Cleavage and Tension Shear Methods inoluded several test specimen designs. The above-mentioned comparisons yielded the following information. (1) mechanical methods proved more acourate than those based upon wood failure estimations, (2) for quality-control purposes the Glueline-Cleavage Test was shown to be equal or superior to the other meohanical methods, and (3) the Glueline-Cleavage Test, when used with specimens designed for research purposes, proved of superior accuracy to a l l others tested. Ad-ditional advantages of the Glueline-Cleavage Method include: (1) the simplest possible test specimen shape and therefore simple and inexpensive manufacture, (2) a maximum yield of specimens from a given plywood sample (a valuable feature with experimental designs requiring large numbers of - 3 -matohed specimens), (3) a lower time requirement per test, (4) much less expensive machinery is required to perform the test, and (5) the exposure of every glue line for inspection purposes when research-type specimens are used* Although the Glueline-Cleavage Test i s believed to be the most aocurate method yet developed, imperfections remain and several further methods have been proposed for increasing i t s aoouraoy. TABLE OF CONTENTS PREFACE Chapters Page I - INTRODUCTION 1 II - THE MECHANICAL TEST VS. PER CENT WOOD FAILURE FOR ESTIMATING BOND QUALITY 5 III - MECHANICAL TESTS OF PLYWOOD GLUE BONDS 1. The Most Desirable Stress Distribution i n a Test Speoimen 16 2. Methods of Reduoing Variability 20 3» Theories Regarding Sources of Excessive Variability i n Test Results 24 4« Theories Regarding the Design of a Better Test Method 27 5 • Summary 30 IV - THE GLUELINE-CLSAVAGE TEST 1* Introduction 32 2. Definition of the Glueline-Cleavage Test 32 3. Exploratory T r i a l 33 4. EGxlO° Speoimen 35 5» Comparison of Glueline-Cleavage, (EGxlO 0), with Tension Shear, Tension Normal, and 40 Block Shear Tests 6. Summary 43 V - ELABORATION OF THE GLUELINE-CLEAVAGE METHOD 1* Preamble 45 2. Auxiliary Tools 46 3» Experimental Design 49 4. Working Plan 50 5. Results 55 (a) Per Cent Wood Failure Vs. Mechanical Test 55 (b) Sensitivity of Designs 57 (o) Glue Strength vs. Wood Strength 62 6. Summary 63 VI - CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE RESEARCH 64 BIBLIOGRAPHY 71 i i i APPENDIX A Table IA - Tension Shear Tests, Springwood bonded to Springwood and Summerwood to Summerwood Table IB - Blook (Compression) Shear Tests, Springwood bonded to Springwood and Summerwood to Summerwood Table 1C - Tension Normal to the Glue Line Plywood Tests, Springwood bonded to Springwood and Summerwood to Summerwood Table 2 - Coefficients of Variation of Breaking Loads -Exploratory Comparisons Table 3 - Comparison of Breaking Loads, etc., Tension Normal to the Glue Line, Tension Shear, Block Shear, and Glueline-Cleavage Tests Table 4 - Statistics of Table 5 Data Figures 1 and l a - The Glueline-Cleavage Maohine Figure 2 - Illustrating the Cutting Plan for "Veneers" Figure 3 - Cutting and Numbering Plan Illustrating the Layup of Glue Blanks for Glueline-Cleavage EGxlO 0 Specimens Figure 4 - Cutting and Numbering Plans for Speoimens Figures 5 and 5A - Tension Normal Specimens Figures 6 and 6A - Tension Shear Speoimens Figures 7 and 7A - Block Shear Speoimens Figures 8 and 8A - Glueline-Cleavage EGxlO 0 Specimens APPENDIX B Table 1 - Randomization of Pressloads Table 2 - Original Data Table 3 - Percentage Reductions i n Strength (or $WF) Table 4A- Comparisons of the Differences i n Predicted Percentage Reduction (in breaking load or per cent wood failure) Design 11 - Design 11% WF i v APPENDIX B (oont'd) Page Table 4B - Comparisons of the Differences i n Predicted Percentage Reductions (in breaking load) Design 11 - Design 4 93 Table 5A - Summary of Slope Ratios 94 Table 5B - Analysis of Variance of Slope Ratios 94 Table 6 - Single Degree Comparisons of Slope Ratios 95 Figure 1 - Cutting Plan for Boards 96 Figure 2 - Designs of Test Specimens 97 Figure 3 - Marking System for Glue Blanks 98 Figure 4 - Marking and Cutting Plans 99 Figure 5 — Logarithmic Transformation; Typical Plots 10G Figure 6 - Representative Plots of Data of each Design for each Treatment 101, 102 Figure 7 - Patterns of Curves for Weathering Treatments I to VI for each Design 103, 104 V PREFACE The researoh forming the basis for this thesis was i n i t i a t e d i n 1950, Chapter IV, The Glueline-Cleavage Test, was the subject of a paper presented before the Sixth National Annual Meeting of the Forest Products Researoh Society at Milwaukee, Wisconsin, i n June, 1952^^^.-^ The author wishes to thank Col. J.H. Jenkins, Mr. R.M. Brown, Mr. E.G. Fensom, Mr. J.B. Alexander, and Mr. W.J. Smith, a l l of the Forest Products laboratories of Canada, for their co-operation i n authorizing the Project Work included i n Chapters TV and V, and for other aid that made possible the oonduot of this researoh. He wishes to thank the members of the Faculty of Forestry, especially Professors R.W. Wellwood, J.W. Ker, J.H.G. Smith, and J.W. Wilson for their valuable assistance. Professor S.W. Nash of the Mathematios Department and Professor W.O. Richmond of the Meohanical Engineering Department of the University of British Columbia, and Mr. D.G. Miller of the Forest Products Laboratories of Canada contributed valued advice. The stenographic work, compiling and ohecking was greatly f a c i l i t a t e d by the help of Mrs. P.D. B i r r e l l , Mr. H.G.M. Colbeok, Mrs. E. Cutforth, Miss W.G. Dyer, and Miss M.L. Wells of the Vancouver Forest Products Laboratory. The work involved i n planning, conducting, and analysing the researoh reported, and especially the preparation of this thesis, has been an invaluable experienoe to the author both personally and professionally i n his work at the Forest Produots Laboratory. It i s hoped that further opportunity w i l l be provided for additional researoh suggested by this study. The figures i n parentheses refer to the bibliography at the end of the text. CHAPTER I - INTRODTJCTIOS 1. The general purpose of testing glue joints i s to measure the strength of the bond* This i s necessary i f step "by step improvements are to be made i n the "effioienoy"^/ of glues and the economy of gluing processes* The art of veneering i t s e l f i s an ancient one, having been prac-tised by the craftsmen of twenty or t h i r t y centuries ago. In spite of the antiquity of i t s origin no real effort was made to investigate the more sci e n t i f i c aspeots of plywood manufacture u n t i l very recently. Although sufficient practical knowledge has been accumulated to allow reasonably satisfactory mass production, manufacturers are oontinually faced with problems about which they have l i t t l e , i f any, fundamental knowledge. The broad reason for plywood research is to provide some of this needed fund-amental knowledge, i n this oase plywood bonding information. One method of appraising bond quality i s to make use of the per-centage of wood failure developed as a result of rupturing the bond. Any broken glue line can be used. To use the method to advantage i t is neces-sary to make the observations on specimens which have been prepared by a ^An "ef f i c i e n t " adhesive i s defined by Knight^ 0' as* One that maintains an adequate bond between the wooden elements under conditions of exposure that the joint has to withstand. The term "adequate bond" has a wide interpretation; i n the small, highly stressed joints of a i r c r a f t or high-powered boats, only the maximum possible strength i s adequate, but for very many purposes a comparatively weak joint gives everything necessary. Whether the joint has high or low strength, i t must be permanent for so long as i t s use demands. Here again, there are degrees of permanence; high-grade furniture i s expeoted to last at least a generation, but i t is sufficient i f non-returnable packaging containers survive the one journey. This useful definition embraoes the multitudinous details of the gluing prooess as well as the adhesive. - 2 -standard!eed procedure. Broken Plywood Glue Shear speoimens^* ^» ^ are one common source of such observations. The Knife Test(®» ^4) method i s another standardized system for estimating percentage wood failure* A seoond type of estimate of bond quality which has found some favour i s the visual examination or de lamination method^* 2 4 ) # Here the prooedure i s to expose the speoimen to the elements, or to some form of accelerated weathering treatment calculated to set up stresses i n the bond* The amount of delamination which develops is used as the oriterion of bond quality* This i s really a mechanical test with the forces sel f -applied by the hygrosoopioity of wood but laoking any measurement of these forces* Neither the visual method nor the percentage wood failure method provides a universally acceptable estimate of the strength of a bond* Eaoh f a i l s , therefore, to meet a fundamental requirement of the perfect bond test* The most universally accepted method of estimating bond quality i s by means of mechanical tests* These may be grouped into four types which have been designed to apply the following stresses to the bondt (1) Tensile (Tension Perpendioular to the Grain Plywood Test(45), Rudkin( 5 8), Glueline-Method( 2 7\ Platow and Tftetz(2\ (2) Shear (e*g* Blook Shear( 1* 40» 41)f Double eompression Shear( 14), Double Tension S h e a r ^ \ (3) Combined tensile and shear (Simple-lap Test P i e o e ^ * ^ \ Plywood Glue ShearU* 40, 4l))# (4) Combined tensile and compressive (Spandau(^» • ^ ) * In the oour3e of routine mechanical testing of glues i t has frequently been necessary to repeat tests because the specimens broke through the wood (100$ wood failure) but at loads less than the allowable wr f T r i - i r — T h e s e conditions developed i n testing for Tension Shear in accord-ance with R.C.A.F. Specification C-22-2^7) (using either sugar maple or yellow biroh). The vari a b i l i t y in test results appeared to be exoessive. and attributable to the wood or method of test rather than to the adhesive-/ under test. When studies were undertaken regarding the gluing properties of Douglas f i r veneers, using either the Plywood Glue Shear T e s t ^ or the "Tension normal to the Glue Line Plywood T e s t " ^ ) , the problem appeared to be even more acute. Dissatisfaction with the results obtained when using standard test methods, and a continuing demand for a means of appraising the quality of glued joints, made i t advisable to investigate the problems pertaining to the testing of plywood bonding. Most of the plywood produced i n Br i t i s h Columbia i s manufactured from Douglas f i r . A l l of the Douglas f i r and western softwood plywood manufactured i n this provinoe i s bonded with hot-press phenolic resin ad-hesives and i s marketed as "weatherproof" grade. In spite of the faot that Douglas f i r plywood has been successfully manufactured for more than twenty years there are s t i l l many problems requiring investigation. One of these and one whioh i t has been the author's privilege to study, i s the effect of different dryer temperatures upon the bond quality. Because certain problems of testing are different when researoh i s limited to one speoies of wood rather than a ohoioe of woods and for the reasons discussed above i t was decided to limit a l l testing to Douglas f i r and to hot-press phen-o l i c resin adhesives. In this thesis the words adhesive and glue are used synonymously. - 4 -This chapter has been devoted to a brief explanation of the purpose of testing i n general and of testing plywood bonds i n particular. A short description of methods of estimating bond quality and some of their deficiencies has been included. Reasons have been presented for limiting the research to phenolic resin bonds of Douglas f i r veneers. The primary objeotive has been set to develop a better method for estimat-ing the quality of plywood bonds. - 5 -CHAPTER II - THE MECHANICAL TEST VS. PER CENT WOOD FAILURE FOR ESTIMATING  BOND QUALITY For the purpose of this thesis bond quality w i l l be defined as the maximum stress, which may be imposed upon the bond without causing i t s failure or rupture of the adjoining wood. It ia the effective strength of the bond at any ohosen instant, i.e., the maximum stress which the bond was capable of withstanding, minus internal stresses, fatigue stresses, and reductions i n the strength properties of the wood. The quality of the bond between two pieces of wood w i l l vary with any treatment to which i t i s subjeoted, such as ohanges i n moisture content, or i n temperature, or by ohendcal or fungal attack, A mechanical test (of a wood-glue bond) i s one which measures the bond quality by means of a "breaking load" that i s related to the ultimate stress which the bond or the wood withstood just before failure occurred, Acoording to the literature mechanical testing of adhesives i s of very recent origin although gluing i t s e l f dates back many oenturies. The earliest reoord of mechanical tests n o t e d i s reproduced, i n part, because of i t s reference to earlier work and i t s generally applicable back-ground. This was the f i r s t of three very comprehensive reports issued by the (British) Adhesives Research Committee between 1922 and 1932. 10, An examination of the scattered literature of the sub-ject quickly shows the unsatisfactory state of affa i r s as regards the testing of adhesives by means of mechanical tests of cemented wood joints. There i s no standardised method of carrying out such tests, and i t i s evident that muoh investigation must be undertaken before there can be adopted any generally acceptable standardised prooedure. Thus, among the variable faotors i n such tests are the nature of the tests themselves (whether tension, shear or impact), the nature and oondition of the wood used, the method of preparing the adhesive and the wood for the test joints, and the form and size of joint employed i n making up the test pieoe. - 6 -11. No r e a l l y satisfactory methods of testing glues, etc., mechanically, otherwise than by means of wood joints, appear to have been devised. Weidenbusoh (1859) employed rods made up of plaster of Paris and glue, but the results appear to "be uncertain. G i l l (Journ. Ind. Eng. Chem., 1915, 7# 102) used briquettes made up of fuller»s earth, diatomaoeous earth, quartz sand or sawdust with various glues, but obtained similarly unsatisfactory results. The same investigator employed a modification of Setterberg ,s method (Sohwed. Teoh. Tidskrift, 28, 52) i n which the strength of glued paper is measured; the results of this test are of qualitative, rather than quantitative, significance. G i l l (loo. cit. ) has also used porcelain, glass and t i l i n g i n the making of glued test joints, but has found them a l l unsatisfactory. Hopp (Journ. Indust. Eng. Chem., 1920, 12, 356) used strips of dried glue ground to standard size, and tested them i n a Sohopper machine. He states that the results of his tests were oondord-ant, but i t is very doubtful whether the tensile strength of a dried sample of glue i s s t r i c t l y indicative of the behaviour of that sample of glue when used i n making a joint between two pieces of wood In the ordinary way. It would seem to be generally agreed that i n order to test the value of an adhesive for use i n jointing timber, the most useful information i s to be obtained from wood joints made up with that adhesive. 12. At the outset of their work i t seemed to the original Committee that while the existing meohanical tests of adheslves for timber were sufficiently good to enable high-quality ad-hesives to be differentiated from poor ones, i t would be d i f -f i c u l t to make step-by-step improvements i n good or bad glues unless mechanical tests were devised which would make i t pos-sible to distinguish with oertainty between samples showing moderate differences of merit. With this objeot i n view invest-igations were undertaken by Major A. Robertson, R.A.F., at the Royal Airoraft Establishment, and a report was submitted by him to the Committee early In 19l9. This report has sinoe been issued* by the Aeronautical Research Committee, and i t w i l l here suffioe b r i e f l y to indicate i t s nature. 13. The various tests whioh, up to that time, had been employed for the determination of the strength of oemented joints i n timber were examined and were each found to be i n -adequate i n some particular direotion. Tension tests are of two classes:- (a) Those of joints made along the grain; (b) those of end grain joints. In connection with the former "Report on the Materials of Construction used i n Airoraft"and Airoraft Engines", Chap. I I , page 132. Published by H.M. Stationery Office, price 21s. net. - 7 -class, two types of test piece formerly used at the Royal Airc r a f t Establishment were investigated and found to be unsatisfactory on account of the unequal distribution of stress produoed by the bending of the specimen when loaded* Test pieces of two other designs (described i n the report) were examined and found to be satisfactory i n securing a . reasonably good stress distribution. It i s interesting here to note that from some tension tests on a very small piece of glue (turned up from a pieoe of cake glue) the tension stress of the glue i t s e l f appeared to be at least 3,000 lbs. per square inoh, i . e . , more than twioe what i s obtainable normally between glue and wood. As regards the second class of joints—end grain joints—at convenient form of test piece i s described i n the report. Preliminary tests on this type gave promising results, and i t was found that good results could be obtained with soft timbers as well as with hard. Reference i s also made i n the report to the Spandau test (used i n Germany), which i s a modification of the direct tension test; i n this test the pieces are glued end grain to end grain .and the speoimen i s broken by bending. An advantage of the Spandau test piece i s that this oan be used for an impaot test which i t may ultimately be desirable to introduce. 14. Major Robertson's report further deals with shear tests, and an examination i s made of the various types of test pieces which have been used by different experimenters. It i s shown that the majority of these are unsatisfactory for i n none i s the stress on the joint a simple shear. It i s suggested that the easiest metnod of carrying out a shear test i s to make the joint inclined at about 15 or 20 degress to the axis of a tension test specimen (figured i n the report). The stresses i n the joint are then a shear and a tension and, i f the angle i s between the limits given above, the stress causing failure w i l l generally be the shear stress. 15. The report concludes with the outline of a procedure for testing adhesives, based on the experience gained i n the work mentioned. The above gives the barest synopsis of the report, which should be oonsulted i n i t s original form by those interested i n strength tests of adhesives for timber. 16. Since the issue of Major Robertson's report, the attention of the Committee has been directed to other investi-gations, and the report consequently represents the stage to which the enquiry has been brought; i t i s hoped to return to the subjeot at a later date. From various r e f e r e n c e s 1 8 , 42) ^ a p p e a r s that the f i r s t serious testing of adhesives was done during World War I to meet the war-time demands of the (wooden) ai r c r a f t industry. During this period meoh-- 8 -anioal tests of glued joints appear to have been developed independently by the Royal Airoraft Establishment at Farriborough, and the United States Forest Products Laboratory, Madison, Wisoonsin, eaoh having developed test speoimens of different design* The two designs developed by the U.S. Forest Products Laboratory, known as the "Block-Shear" and the "Plywood-Shear" tests, have been des-cribed by several authors^* 40, 4 l ) # Some eight or more designs of test specimen were investigated by the (British) Adhesives Research Coinmitfcee between 1920 and 1932^1-'» ^ ) # o r m o r e 0 f these had been used during World War I for purposes of glue specification* After an extensive investigation two designs were recommended as the best of those considered* One, the "Simple Lap Test-Pieoe" was designed for shear test3, and the other, the "Spandau Test Piece" for testing adhesives i n t e n s i o n ^ * ^ ) # Under the pressure of the Fi r s t World War great strides were made in both practioal and theoretioal fields of adhesion i n America and Britain* The researches carried out at this time by such men as Anderson, Browne, Brouse, Hopkins, Lee, McBain, Robertson and many others were thor-ough* These workers were f u l l y aware of the imperfeotions, as well as the merits, of their (mechanical) test methods. They realized that breaking loads are not stresses, that every change i n the dimensions of a test specimen altered the breaking load, and after taking every precaution to standardize conditions and procedures a very large variation would remain between their test results. In faot their work was so sound that their recommendations have formed the basis of most consequent adhesive testing* In 1945 Platow and Dietz^ 2) presented another specimen design for testing bonds i n tension. A new approach was set forth, that of del-iberately introducing stress concentrations i n the glue line instead of attempting to eliminate them* The wood-to-wood speoimen has been designed to isolate glue failures from wood .failures by incorporation of a notch to localize stresses.^) In recent years several new mechanical test methods have been developed. Wakefield^) desoribed a "Tension Normal to the Glue Line Plywood Test" designed "bo give a positive and quantitative value to the adhesion between plies that can be duplicated within normal experimental error". Rudkin^^) presented "A Simple Method of Testing Glue Lines i n Tension", whereas Elmendorf(17) used the "Torsion Shear Test" for plywood. More reoently Laoey and Howe^2?) have desoribed a "Glueline Method", employ-ing a wedge to oleave a speoimen along the glue line, while Elmendorf has introduced a variation of the "Cantilever" type^^) developed by Robert-son for the Adhesives Researoh Committee. The work of the Adhesives Researoh Committee^'* 15) &i\ based upon breaking loads obtained from meohanioal tests of glued joints. Wood failure was a nuisance which they t r i e d to minimize i n their attempts to measure the strength of (strong) glues by means of wooden test pieoes. It was quite clear to these workers that their interest lay i n the strength of the bond and their objeotive was to devise means of measuring this bond strength. One of the deciding factors i n their choice of a standard test piece was as followst It was found that i f the simple-lap test pieoe were assembled with speoifio attention to grain direotion, gross timber (wood) failure could be eliminated. To obtain a minimum of timber (wood) fai l u r e , the timber should be out so that the grain makes a small angle with the test pieoe. In their researches no reoord has been presented of the per cent wood f a i l -ure ooourring i n the broken specimens. - 10 -The earliest U.S. Forest Products Laboratory r e p o r t ^ ) whioh i t has been the author's privilege to read, when referring to the Plywood Shear Test results, states: "The glued surface must not f a i l at a load of less than 150 pounds per square inch". No mention i s made of per oent wood failure.. Another Forest Produots Laboratory report(^l) states: For eaoh (block shear) specimen tested, notation i s made of the breaking load and the estimated percentage of the glue line area i n whioh the wood splinters. Two or more duplicate joints are usually prepared, each one giving 10 speoimens for test. The average and the minimum breaking load and the average per-centage of wood failure are generally taken as the f i n a l reoord of the test. No mention i s made of how muoh weight i s given to each i n evaluating the quality of the bond. Truax (40) has discussed the fallaoy of using wood failure as a means of evaluating the strength of glued joints. It appears that early researchers accepted measurement of bond strength as their objective. The interpretation of results was complicated by a laok of knowledge regarding the stress distributions existing i n speoi-mens at the time of failure, and by the great number of souroes of variab-i l i t y with whioh they had to oontend. It appears that by 1938 serious consideration was being given to the use of per oent wood failure as a oriterion of bond quality, at least for plywood specification purposes i f not for researoh purposes, as witness the following quotation from Perkins (36), Fortunately, the Forest Produots Laboratory did have, (ex-perience with tests for Exterior Type Plywood) from panels of various kinds of plywood whioh had been weathering for nearly five years. It was their conclusions that i f standard plywood shear speoimens revealed 50 per oent wood failure after having been subjected either to three and a half cyoles of cold soaking and drying or to two cycles of boiling and 145°F. drying, that the panel oould be expeoted to have a long l i f e i n service. Just how long, no one was w i l l i n g or able to prediotj ... One of the f i r s t impressions gained i n this study - 11 -was the importance of the amount of wood failure developed when testing the joints. It i s not clear whether or not this use of per cent wood failure as a sole estimate of bond quality developed from a oonviotion that i t was a superior method, or whether i t was put forward as a convenient substitute for break-ing loads with a l l their vagaries. At any rate, by 1950 the Douglas F i r Plywood Association had b u i l t up data whioh tended to confirm their "belief i n a close relationship between wood failure and durability" The breaking loads of their specimens used for per oent wood failure determina-tion do not show appreciable correlation with durability. There are several possible explanations for this laok of correlation. Brouse^) has shown that a change i n the thickness of any veneer from that of the ohosen standard ohanges the anticipated breaking load for that speoimen. These anticipated breaking loads for yellow birch 3-ply plywood varied from 120 pounds for 1/64" face and baok veneers and l / 8 " oore veneer to 740 pounds for l / 8 " faoe and back veneers and l / 6 4 " oore veneer. It appears that this faotor was not kept under control and that plywoods of various thicknesses were included i n the Douglas F i r Plywood Association tests. Bethel and Huffman^6) have shown that the orientation of lathe checks with respect to the saw cuts can make highly significant differences i n the values of breaking loads and per cent wood failures obtained. There i s no indication that this faotor was con-tr o l l e d i n the Douglas Fir Plywood Association work. The Douglas F i r Plywood Association^'^ was satisfied that per oent A wood failure, as read from stanflard Plywood Shear specimens subjected to 4 hours boiling, 20 hours drying, and 4 hours boiling, provides an estimate of the bond quality superior to that obtainable from considering the mag-nitude of the breaking loads. - 12 -In 1949 Laoey and Howe^2"^, i n discussing the ohoioe of a measure of bond quality for researoh purposes, stated: The choice of methods lay i n the f i r s t plaoe between a meohanioal strength test and the visual examination of glue lines that have been s p l i t open* Experience at the Laboratory with veneer-base joints has favoured the visual test as being more sensitive to i n i t i a l degrade, quicker to carry out and independent of special testing f a c i l i t i e s , and more convenient for examining large areas and particularly for ready-made joints. No solid wood test piece has, however, been developed for examination i n this way. It seems clear that had they been studying veneer bonding problems they would have used either the visual or the per cent wood failure method i n preference to the results of mechanical tests. According to Knight^ 2^), " i n B r i t i s h specifications, wood failure has no plaoe, although most testers agree that adhering fibre i s desirable". It appears from Newall^ 2) that a move i s afoot to introduce per cent wood failure as an alternative to the meohanioal test. The existenoe of the following statement i n B.S. 1455 (1948)^/, and B.S. 1088 (1951)2/, "by means of a knife test on plywood of any species or thickness, or pieces of any convenient size being tested dry or after appropriate water treatment", would seem to indicate that per oent wood failure , or something olosely akin to i t , definitely has a plaoe In B r i t i s h speoifioations. On the basis of the author's experience this trend, while i t may be justifiable from an adhesive c e r t i f i c a t i o n or production testing stand-point, i s to be deplored from a researoh standpoint. It i s not possible to t e l l i f a bond i s adequate when no attempt i s made to measure i t s strength. It i s conceivable that this strength may be approximated by using other than meohanioal tests, but only after the bond strength has been estimated by the use of these tests. If per cent wood failure is to be used as an estimate 4/ — British Standards Institute Specification. - 13 -of bond strength a correlation must be established between per oent wood failure and meohanioal test data* After such a relationship has been established bond strength could then be predicted from per oent wood f a i l -ure* It i s axiomatic therefore that adequate mechanical tests are fund-amental to a l l (wood) glue and gluing research* These (meohanioal methods) are the basic tools used to evaluate the results of a l l other methods of test, be they physioal, chemical, pathological or combinations of these and others* It i s not intended to imply that to be useful the results of meoh-anioal tests must be reduoed to glue line stresses, although this would be desirable. Breaking loads as measured by mechanical tests, provided they are positively correlated with the actual stresses i n the glue line, may be equally useful for many purposes* A fundamental consideration i s that the strength of the bond i s of major interest, not that of the adhesive i t s e l f . An adhesive whioh develops the f u l l strength of the wood under the required service conditions f u l f i l l s a l l requirements except possibly that i t might be less expensive or more tractable. No more oan be expected of the wood, and attempts to estimate the strength of a strong adhesive through the medium of weak wood (and many of them have been made) have always been doomed to failure* A drawback to the use of per oent wood failure as an absolute measure of bond quality i s the great difference i n the appearance of joints which may be induced by altering the angle, of grain with reference to the direction of application of the load or saw outs. The (British) Simple Lap Test Pieoe(*4» 15) purposely designed to minimize per cent wood failure so that any representative estimate provided by these specimens would be muoh lower than those provided by the (United States) Block Shear Test(l» 42) ^  - 14 -or the (Canadian) Tension Shear Test Pieoe^*^. These objections are min-imized for specification testing of one speoies of wood, one class of adhesive (e.g. phenolic resin adhesives), and one standardized design of test piece. This is the situation with the Douglas F i r Plywood Association, and the use of per oent wood failure as a measure of bond quality may well be just i f i e d on the grounds of expediency. The Douglas F i r Plywood Association, working under conditions that favour the use of per oent wood failure, found i t necessary to teach estim-ators how to "read" per cent wood failure so that data colleoted by different persons, or the same person at different times, w i l l agree within acceptable lim i t s . Their method i s to circulate standardized sets of specimens from estimator to estimator and require that each check his readings against the standard. This standard, although adopted only after careful study by ex-perts i n the wood gluing research f i e l d , i s s t i l l an arbitrary one. Another group of experts would quite l i k e l y have chosen a different standard with the same specimens, and would almost certainly have arrived at a different standard had they used a different design of test piece or different species. If the Douglas Fir Plywood Association has found i t neoessary to take these precautions to coordinate the estimates made by different "readers", how much more elaborate a system is required i f per oent wood failure i s to be used under less suitable conditions. Different test pieces, species, ply-wood and laminated wood, different adhesive classes, to say nothing of estimators from laboratories spread across the world greatly increase the hazards associated with using per cent wood failure as a universal estimate of bond quality. If this i s to be used as a measure of bond quality a world-wide exchange of standardized specimens i s required. Photographio keys or written descriptions have not proven satisfactory substitutes. Other - 15 -references(4» 24» 26, 31) have been helpful i n dealing with per cent wood fa i l u r e . In spite of the defects of mechanical test results, such as excessive variability and inaoourate knowledge of how the breaking loads should be interpreted (due to inacourate knowledge of the stress d i s t r i -butions at the moment of fa i l u r e ) , they s t i l l oome closer to actually measuring the quality of a bond than do other so-called "measures"• At best, other estimates such as per oent wood failure, can only be used as substitutes for the mechanioal test after the two have been shown to be correlated. Only then can the mechanical strength of a bond be predicted from per cent wood failure readings. Furthermore, the defects of the mech-anical tests appear to be more amenable to solution through further research than do those associated with per cent wood fai l u r e ; certainly this i s one of the impressions gained from the research herein reported. When a l l things are taken into consideration i t appears that the results of meohanical tests, i n spite of their imperfections, provide a more universally applicable standard of glue line quality than does per cent wood failu r e . - 16 -CHAPTER III - MECHANICAL TESTS OF PLYWOOD GLUE BONDS 1. THE MOST DESIRABLE STRESS DISTRIBUTION IN A TEST SPECIMEN It i s interesting to note that when meohanioal testing of ad-hesives was undertaken by the Adhesives Researoh Committee (-^ ) the aims appear to have been to develop test speoimens, and methods of testing them, whioh would apply uniform shear or uniform tensile stresses over the entire area of the bond i n order to obtain a true measure of the shear or tensile strength of the bond. The authors of the F i r s t R e p o r t ^ o f that Committee had this to say i n reference to tension tests made along the graint Two types of test pieoe formerly used at the Royal Airoraft Establishment were investigated and found to be unsatisfactory on account of the unequal distribution of stress produced by  the bending of the specimen when loaded. In reference to shear tests i Major Robertson's report further deals with shear tests, and an examination i s made of the various types of test pieoes whioh have been used by different experimenters. It i s shown that the majority of these are unsatisfactory for i n none i s  the stress on the joint a simple shear. It i s clear that their objective was to obtain uniformly dis-tributed (pure shear or pure tensile) stresses over the test area. It i s equally clear that they were not satisfied with the results obtained. This approach i s i n contrast to the views of some of today's investigators. When discussing different test pieoes^4) the Committee says» It has at various times been suggested that glued joints should be tested under stress conditions approaching as close as possible to pure shear i n the plane of the joint. In a l l attempts to obtain a pure shear stress i n practice, bending or distortion of the parts has introduced secondary stresses whioh have probably influenced the actual failures. Sinoe, however, suoh an effeot usually occurs i n the failure of actual glue joints i n servioe, there does not appear to be any logical reason why adhesives should be tested by any close approximation of pure shear. Both test pieoes which were eventually recommended by the Com-mittee^S), the Simple Lap Test Pieoe (Tension-shear) and the Spandau Test Pieoe (end-grain joint tested i n bending), were types which did not give even a close approximation to pure shear or pure tensile stresses. Data are given^5) to show that breaking loads of simple lap (tension-shear) specimens of different areas are not constant when reduoed to pounds per square inch of glue line tested, but vary from 1,252 p . s . i . for a one-inoh lap to 2,334 p . s . i . for a one-quarter-inoh lap. With this type of speoi-men, therefore, dividing the breaking load by the oross section of the glue line does not give a r e a l i s t i c measure of the magnitude of the stresses to which the glue line has been subjeoted. The results of this type of test should be recorded as breaking loads i n pounds rather than as pounds per square inch, at least u n t i l an acceptable stress analysis is performed. Only then may the results be reported as maximum stress on a unit basis. What constitutes the most desirable stress distribution i n a test speoimen? The following quotations from Platow and D i e t z ^ indicates that this problem was s t i l l not resolved as reoently as 1945. Strength testing of adhesives i s a peculiar problem because i n any useful test specimen the adhesive being tested i s only a minor part of an assemblage, and the only way to exert any foroe upon the adhesive i s through the material bonded, or substrate. Ho feasible way of grasping the adhesive directly has been found. Complicating the problem i s limited knowledge of the nature of adhesion. .Although i t appears to be generally accepted that adhesion i s primarily an actual adherence of the ad-hesive to the surface of the material bonded and not some mechanical keying or anohoring phenomenon, no matter whether the surfaces be porous or perfectly smooth, l i t t l e is known about the forces, moleoular or otherwise, which cause the adhesive and the substrate to cling to each other. Con-sequently, i n a l l strength tests a considerable degree of - 18 -uncertainty exists as to what i s being tested* A further complication i s that i n any strength test of ad-hesives at least two strength factors are being tested simultaneouslyt the strength of adhesion of adhesive to the adjacent surface, and the strength of oohesion of the adhesive i t s e l f . Since the only feasible way to exert foroe upon the adhesive is through the substrate, i t follows that the substrate i s being tested at the same time as the strengths of adhesion and cohesion* The result i s three simultaneous strength tests, with the ohanoes often more than even that failure w i l l not ocour i n the adhesive at a l l * Keeping these complications i n mind, there are two principal approaches to the problem of devising strength tests for adhesives. One, whioh may be called the 'purist' or aoademio approach, i s to devise tests i n whioh only pure stress of one kind—for example, tension or shear—is exerted on the adhesive, and at the same time care i s taken to assure failure i n the adhesive, i n either adhesion or oohesion, and not i n the sub-strate. A second approaoh.is the more pragmatic one of using a test specimen relatively easy to make; testing i t i n as simple a manner as possible consistent with obtaining more or less pure stresses; and taking the viewpoint that i f failure actually oocurs i n the bonded material, rather than i n the adhesive, i t proves that the adhesive is the stronger, there-fore has successfully passed the test, and no more need be asked. If failure occurs i n the adhesive, so much the better, since numbers oan be reported representing strength values under the conditions of test. There are objections to both, procedures. While perhaps fund-amentally sound, the purist's approach i s d i f f i c u l t and may easily require rather elaborate equipment as well as complicated test speoimens to assure pure stress. Furthermore, with strong adhesive and oohesive forces and weak substrates, i t may be almost impossible to obtain failures i n the adhesive i t s e l f without so altering the conditions as to destroy their praotical significance. The more pragmatio approaoh i s open to the criticism that i f combinations of stress are present i n a test, there i s no sure way of determining whioh stress was mainly responsible for the failure and therefore there i s no good way of trans-lating the results into pure strength values for the use of designers. The oonsequenoe is to force the employment of a great many different kinds of test specimens to reflect the different types of assemblage that might be found i n actual use. A point i n favour of suoh a procedure i s , of course, the faot that i n any complex struoture the designing engineer seldom i s certain of the exact distribution of various kinds of stress i n a joint and that i t may for that reason be neces-sary to test any given joint to determine i t s strength under - 19 -oauditions simulating actual use* On the other hand, the designer could approach his design with more confidence i f he were sure that the strength values he was using actually represented those strengths uncomplicated "by other factors™/. These arguments help to foous the reader's attention upon the problems involved but seem to confuse rather than olarify the issue* As far as a designer i s oonoerned, a l l he requires is that the bond "develop the f u l l strength of the wood"^/ under the required conditions of servioe* Once assured of this he has no further interest i n the adhesive but uses the strength properties of the wood for his purposes* Inasmuoh as there are many adhesives of this caliber available there seems to be l i t t l e need for pure stress values. Platow and Dietz (2) appear to have resorted to the "more prag-matio" approaoh because they designed a test speoimen to deliberately concentrate the maximum stresses i n the glue l i n e . ... The wood-to-wood speoimen has been designed to isolate glue failures from wood failures by the incorporation of a notoh to localize stresses^ 2'. It i s interesting to note that photoelastio analyses^ 2* ^ » ^9) indicate that the stress distribution over a seotion subjected to a "pure" stress i s not uniform across the seotion. The purist's dream of achieving a uniformly distributed pure tensile, shear, or compressive stress was, i t appears, foredoomed to f a i l u r e . The aotual distribution of stresses i n a specific wood-to-wood bond w i l l probably never be known due to the number of contributing faotors among which may be mentionedt — — — — Phrases suoh as this or (referring to the strength of a glue) "weaker than the wood", "stronger than the wood", etc., w i l l be found i n the text. It i s realized.that these are inexact statements of the existing oondition but they express the idea more clearly than a more involved statement would. - 20 -(a) the stress-strain characteristics of the wood, (b) stress concentrations existing i n the wood before gluing, ( 0 ) the stress-strain characteristics of the adhesive, (d) stresses induced i n the adhesive by shrinking or swelling when during, (e) stresses induced i n the joint by swelling of the wood due to moisture absorbed from the adhesive, (f) stresses induced i n the joint by swelling or shrinking due to moisture picked up or lost by the wood after being bonded and before being tested, the most severe stresses probably being induced by soaking or boiling the specimens prior to test, (g) stress concentrations induced i n the wood due to differential swelling or shrinking of springwood and summerwood with ohanges i n moisture content of the wood, (h) the pattern of stresses induced during testing of the specimen; each method of test, size of specimen, speoies of wood, e t c , w i l l induce a different pattern, and (1) the effect of the moisture content of the wood upon i t s stress-strain characteristics. In spite of the apparent impossibility of obtaining aoourate knowledge of the stress distribution i n individual specimens at the moment of failure there i s , nevertheless, some hope of reducing the v a r i a b i l i t y i n test results through stress analyses. The accuracy of estimates based on these results can be improved by more intensive study of this stress dis-tribution problem. 2 . METHODS OF REDUCING VARIABILITY Our principal objects were f i r s t to discover i f possible the causes of the wide variations whioh occur between the results - 21 -of individual tests of the same glue when any of the usual form of test pieoe i s employed and then to develop a form of test pieoe from which more concordant results might be ob-tained* This quotation, from the Second Report of the Adhesive Researoh Committee(H), could well have been used i n the foreword of this thesis. The problems s t i l l remain. The sources of variation were investigated rather exhaustively under the Committee's auspices, and complete instruc-tions were prepared for their minimization under the following headingsj I. Timber (a) Moisture content. (b) Structure, etc. (c) Mechanical preparation of the surface. (d) F i t of glued surfaoes. (e) Inclination of grain. (f) Temperature at -time of gluing. II . Glue (g) I n i t i a l variations i n glue. (h) Previous history. (i) Proportions of constituents, (j) Temperature of heating. (k) Duration of heating. III. Operation of gluing (1) Method of application. (m) Room temperature and humidity. (n) Delay i n bringing surfaoes i n contact. (o) Clamp pressure. (p) Time under pressure. (q) Aoouracy of pressure j i g . (r) Rate of cooling. (s) Glue f i l l e t . (t) Temperature and humidity of .conditioning chamber, (u) Time of conditioning. IV. Testing (v) Acouraoy of alignment of specimen. (w) Loading grips. (x) Temperature. (y) Humidity. (z) Rate of loading. - 22 -This work was so well done that few, i f any, sources of variation were overlooked. Other investigations have oonfirmed the conclusions regarding these sources of v a r i a b i l i t y 4 2 ) # Having done their utmost to minimize the v a r i a b i l i t y attributable to these sources the authors^^) s t i l l lament that, One of the most unsatisfactory features associated with a l l present methods of test i s the large variation which occurs, the cause of whioh has not yet been traoed. In any one group of tests the difference between the maximum and minimum values i s usually of the order of 20 to 30 per oent of the arithmetic mean. Claims for smaller variations should not be accepted without investigation, as the anomalous position exists that i f the tests are suoh as to allow timber failure, the apparent variation i s frequently reduced, the strength of timber being usually more constant than the apparent strength of the glued joint. Variations appear to exist, not only among joints from any one group, but also among groups which have been prepared under apparently similar conditions. This increases the advisability of using a large number of test pieces i n each group to give greater accuracy i n the mean result, but owing to experimental d i f f i c u l t i e s (provision of jigs, etc.) the number composing each group i s very limited. It should therefore be remembered that i n many of the experiments here described, the probable aoouraoy of the mean results i s i n -sufficient to allow of definite conclusions being formed, and although the tests are useful i n indicating the probable trend, i n particular oases the indication may be erroneous. To assist the estimation of the amount of reliance whioh may be placed on the results, the mean errors have been calculated using Peter's approximation. These have been given i n two forms: (a) Plus and minus limits on the arithmetic mean. (These give the limits within which the true result i s as l i k e l y to l i e as outside). (b) The probable percentage error of a single observation whioh gives some guide to the relative degree of variation which i s ooourring among groups containing similar numbers of results. The situation i s not greatly ohanged except i n one important - 23 -respectj the science of statis t i c s has been developed to assist i n analyz-ing variable data suoh as the test results of glued joints. To this can be added the inoreased knowledge regarding distribution of stresses i n stressed structures. The theory of s t a t i s t i c s , especially Correlation Analysis and Analysis of Variance techniques, provides one of the most, powerful tools at the disposal of the researcher. S t a t i s t i c a l methods offer a means of segregating controlled variation induced by treatments from the uncontrolled or "error" varianoe on a mathematical basis. Early workers laoked these methods except i n the very primitive form mentioned i n the above quotation. The distribution of stresses i n a specimen is obtainable through the use of either photoelastic or complicated mathematical analyses. Certain of these analyses have dispelled any suggestion that stresses act uniformly over the test area. The influenoe of lathe check orientation on Plywood Shear Test Results (6) i s another source of variation for whioh a method of reducing v a r i a b i l i t y has been introduced only reoently. Another aid i n the campaign to reduce v a r i a b i l i t y i s the use of the ratio breaking load of glue line instead of a breaking load. Orth(35) breaking load of the wood used a method applicable with the Blook Shear Test. Marra and Wilson^^) have used the same principle for what they c a l l a "gluability index". Selbo and Olson(39) employed the same principle for presenting data on a variety of joint types. This principle, applied to plywood, has been used i n the research recorded i n this thesis. A further method of increasing the aocuraoy of mechanical test interpretation concerns the fallacy of attempting to t e s t a strong adhesive through the medium of weak wood. Much research effort has undoubtedly been wasted because the worker attempted to measure the strength of strong glues - 24 -by means of weak wood. In order to minimize recurrence of this d i f f i c u l t y the following method i s proposed i n contrast to the normal praotioe of using an adhesive only as recommended by the manufacturer. It i s proposed, that several formulae of this adhesive be used. These additional mixtures deliberately include some "weaker than the wood" and, when used i n conjunc-tion with the method suggested i n the next paragraph, the pattern of test results w i l l indicate which of the adhesive formulae, i f any, form bonds "as strong as the wood". This i s the information required to avoid the fruitle s s task of trying to draw conclusions regarding the strength of an adhesive whioh i s "stronger than the wood". The following i s another method whioh should add to the accuracy of predictions based on the results of mechanical tests. The prinoiple i s to base the judgment of bond quality upon the rate of strength reduction induced by a series of "accelerated weathering" oycles of increasing sever-i t y , i . e . , upon the rate at which the strength of the bond i s reduced rather than upon the absolute strength. 3. THEORIES REGARDING SOURCES OP EXCESSIVE VARIABILITY IN TEST RESULTS A review of the inherent defeots of mechanical methods was re-stricted to a few commonly accepted tests: Tension Shear, Tension Normal to the Glue Line, and Block Shear. This re s t r i c t i o n was imposed for two reasons: f i r s t l y , the author has had some experience with these particular tests, and secondly, most other types already have been discarded by com-petent authorities. The f i r s t step was to consider i n some detail a l l the factors whioh might be responsible for the great variability of the test results, only the more important of which w i l l be discussed. - 25 -In s t a t i s t i c a l theory, the Total Variation occurring i n a popula-tion of test results is the sum of a l l the individual sources of Variation. In this instance the Total Variation is the sum of those inherent i n the wood, the design of the test speoimen, the adhesive, the testing machine, the workmanship, and f i n a l l y that deliberately introduced for the purpose of study. These are commonly segregated into two classes, one attributed to "treatments" and the other to "error". The ideal test specimen and method of testing should be such that a l l variation, except that purposely introduced for study, w i l l consequently be attributable to the wood and the adhesive and none to the method of test, workmanship, or other oontrollable factors. Many broken test specimens were examined for the purpose of locating sources of variation which would lend themselves to corrective action. The more important of the theories con-sidered w i l l be described b r i e f l y . (A) Tension Normal to the Glue Line Plywood Test(45) In this method of test the load is applied perpendicular to the glue line so that a l l sections par a l l e l to the glue line are, theoretically, equally stressed. I f wood were a homogeneous material and workmanship perfect this would probably be true. There are, however, suoh factors as differences i n the physioal and mechanical properties of springwood and summerwood, imperfections suoh as minute seasoning cheoks, manufacturing defects, and the addition of two extra glue lines. It seems inevitable that a stress concentration w i l l be created at some section i n the test pieoe other than the bond under study, and that failure at that point w i l l result except when the bond i s very weak. This was illustrated by the broken speoimens, only a minority of which had broken i n the glue line or i n i t s immediate vi o i n i t y . - 26 -(B) Tension Shear Test^ 9* 1 0 » 5 7 » 42) In this test method the specimen is designed so that stress con-centrations are set up at certain seotions, notably at the base of the saw-cuts. Depending upon the quality of the workmanship this,may or may not be at the glue line where a stress concentration would be of value. It was reasoned that this causes the wood of many speoimens to f a i l f i r s t i n tension, followed by oleavage, whereas the objective i s to test the joints i n shear. It was reasoned further that, apart from the design of the test specimen, two factors would tend to introduce undesirable v a r i a b i l i t y into the test results. These factors are slight differences i n the depth of the saw outs and the pattern of springwood and summerwood at the gluing faoe. For glue aooeptanoe tests these factors have been minimized by using the most suitable speoies; for production testing they have been accepted as inevitable conditions of the test; but for researoh into the gluing properties of a particular species the issue oannot be evaded by substituting another species, neither is i t desirable to aooept these con-ditions as inevitable. (C) Block Shear Test^ 1' 4 0 * 4 2 ) In this test method the foroe i s applied so as to produce a (compressive) shearing stress i n the plane of the bond. This method as described^ 1* 40, 42) probably comes as close as any to producing a uniformly distributed (shear) stress over the joint area, the ideal sought by early workers (14). When an attempt was made to use the method with specimens modified to suit plywood the wood failed i n compression parallel to the grain before the bond was ruptured. This made i t neoessary to glue blocks of wood to the specimen (see Figures 7 and 7A. Appendix A) before the joint - 27 -oould be sheared, 4. THEORIES REGARDING THE DESIGN OF A BETTER TEST METHOD The major requirement of a good test method i s that the results shall be reproducible. With bonds of uniform quality the method of test with least v a r i a b i l i t y should be the best. With perfectly uniform materials the ideal test would always give the same breaking load. It i s axiomatic that the results of a test ishould be expressed i n universally acceptable and unequivooal terms. Unit stresses or breaking loads i n pounds for a standardized test specimen are reasonable examples. Probably the next most important requirement is the simplicity of the test speoimen. A specimen whioh is d i f f i c u l t to prepare f a l l s short of perfection. In addition, the cost of the equipment required to prepare and test the specimens should not be excessive. Finally, i t i s desirable, though not essential, that a test method be adaptable for either researoh or production testing. To meet the requirement that tests of uniform quality bonds shall exhibit minimum v a r i a b i l i t y the method would be expeoted to incorporate as many as possible of the following features: 1. The load should be applied i n suoh a manner that the c r i t i c a l stress w i l l be concentrated i n the glue line whose quality is to be est-imated. The region of maximum stress should be known. A stress system which i s subjeot to mathematical analysis would be an asset. There are at least two methods whioh may be used to achieve this objeotive. The f i r s t has to do with the shape and method of loading the speoimen, i.e., by means of a tensile, compressive, or shearing force, or wedging action. The latter u t i l i z e s the differences i n strength properties of wood i n the longitudinal, radi a l , and tangential directions to concentrate the stresses i n the desired portion of the bond. - 28 -2, In the oase of softwoods the normal variation of springwood and summerwood pattern at a glue line must introduce some unnecessary-va r i a b i l i t y into the test results. It is reasoned that oross-banding edge-grain veneers would yie l d specimens with substantially the same per-centages of springwood and summerwood i n the glue line of each speoimen and thus tend to minimize v a r i a b i l i t y from this source. 3, When veneers are lathe-out from a log, irregularities i n the grain of the wood result i n oertain cells being cut longitudinally whereas others are out at an angle. Those out at an angle afford an opportunity for glue to be drawn into the lumen by capillary action, or to be forced i n by pressure; the others do not. It seems logical that by cutting veneers at an angle to the grain to expose the lumen of most c e l l s , var-i a b i l i t y from this souroe would be reduoed. 4. The surfaces of veneers, whether sliced, peeled, or sawed, suffer damage to some extent during manufacture. The surfaoe damage, con-sisting of oheoks, uneven thiokness, torn splinters, or torn fibres, must add to the va r i a b i l i t y of test results. In order to minimize v a r i a b i l i t y from this source, i t was considered desirable to saw and plane veneers whioh were to be U3ed i n studying the gluing properties of a species. 5. The f i n a l method of providing for minimum va r i a b i l i t y between test results i s standardization. Standardization i n this connection means that the dimensions of the particular source are defined exactly, so var-i a b i l i t y from that souroe w i l l be minimized. It Is known for instanoe, that any change i n the dimensions of a test speoimen, or of the veneers of whioh i t is composed, alters the magnitude of the breaking load. This being so i t i s necessary to specify i n detail the veneer and speoimen dimensions. Another example would be specification of the exaot moisture - 29 -content at whioh speoimens shall he tested. The same principle applies throughout the gamut of sources of v a r i a b i l i t y . Probably there i s not a single feature conoerning a method of test which oan be varied from the chosen standard without introducing some consequent change i n the breaking load. Standards must be set and adhered to f o r suoh widely divergent features as the shape and dimensions of the test speoimen, the details of the machines and accessories used to test the specimen, the details of wood and glue preparation, the details of the bonding process and the moisture content of the wood. A rather complete l i s t of these factors i s included at the beginning of Chapter III, Part 2, "Methods of Reducing Variability". 6. Every precaution should be exercised to assure fine workman-ship. The test specimen should be as simple as possible to manufacture. In hot-press bonding with veneers, single speoimens are more d i f f i c u l t to prepare than sheets from whioh test specimens are out. Sawcuts whioh must touch the glue line are d i f f i o u l t to make accurately. The more complicated the specimen shape the more skilled the labour required to prepare specimens to the required high standard of acouracy. A rectangular solid (box) i s probably the specimen shape most simple to prepare to a high standard of dimensional accuracy. Certain types of specimen, such as the Tension Shear, require three or more times the area of wood actually tested, i n order to apply the load. Where the entire speoimen i s tested, three times as many speoimens oan be prepared from the same quantity of wood. This can be a worthwhile feature when large numbers of matohed specimens are required for a par-ticular design. The cost of a testing machine may not be of major consequence for the large researoh organization but for many smaller establishments this - 30 -feature alone presents a c r i t i c a l r e s t r i c t i o n of methods* An aocurate method suitable for production testing and employing an inexpensive apparatus would be of value to industry* It would generate more interest i n quality control of the gluing process which would lead to higher standards with smaller losses from substandard produots. 5. SUMMARY In summary, the evidence provided by photoelastic analyses^ 2* ^» ^ \ stress analyses^ 9' JS)$ and experimental evidence^ 2* 14, 15, 21) t con-firms the theory that the most desirable stress distribution i s one which i s concentrated i n the glue line . The most desirable test specimen, there-fore, would be one in which the maximum stresses are concentrated i n the glue l i n e . Methods of reducing v a r i a b i l i t y i n meohanioal test results, and the interpretation of these results, include the following: the methods stressed by the Adhesives Research C o m m i t t e e ^ 1 4 , 15) f the United States Forest Produots Laboratory^* 42), lathe-check orientation^), reducing breaking loads to percentages of wood strength(30, 35, 39), using a range of adhesive strengths from "weaker than the wood" to "stronger than the wood" i n combination with the rate of reduotion i n breaking load induced by a series of weathering oyoles of increasing severity, and s t a t i s t i c a l tech-niques such as correlation analysis, slope ratio assays., and analysis of varianoe. The features to be kept i n mind when designing a method of testing bond quality for research purposes include the following: (A) TO minimize va r i a b i l i t y among test results: (1) Concentrate the stress i n the glue line (a) by means of the shape of the specimen and method of - 31 -applying the load, (b) by oontrol of the fibre orientation of the wood. (2) Use edge-grain veneers and oross the springwood-summerwood bands when gluing. (3) Prepare veneers so that the wood oells are out at a slight angle• (4) Use planed rather than peeled or slioed veneers. (5) Standardize the dimensions of the test specimen and the manu-facturing and testing procedure to be followed, using the l i s t s at the beginning of seotion 2 of this chapter as a guide. (6) Demand the highest quality of workmanship. (B) Use the simplest shape of speoimen consistent with loading require-ments. (6) Keep the cost of equipment as low as i s consistent with the neoes-sarily high standard of aocuraoy required. (D) A desirable but not essential feature would be that the method be suitable for production testing as well as for researoh purposes. - 32 -CHAPTER IV - THE GLUELIHE-CLEAVAGE TEST 1. INTRODUCTION The unsatisfactory state of fundamental knowledge regarding the gluing prooess has been presented, followed by a comparison of visual and mechanical methods of estimating glue line quality. The theory of the superiority of the latter technique was then propounded. Deficiencies of certain meohanical methods of evaluating bond quality have been tabulated, leading to the conclusion that whereas a meohanioal method i s desirable, none existing approaches the desired standard of accuracy. Requirements of the ideal method have been reviewed. 2. DEFINITION OF THE GLUELINE-CLEAVAGE TEST The Glueline-Cleavage Test has been designed to meet this need for a more accurate method of appraising the strength of wood to wood bonds. It i s made by placing a knife-edge along the glue line of the speoimen and measuring the force required for cleavage. Figures IA, Appendix A. show the knife and specimen i n position for testing. Figure 1 of the same appendix shows the hydraulio jack which has been adapted to apply the load, and the gauge which reoords the foroe applied. Any alternative method of aoourately measuring the applied load would be suitable. I n i t i a l l y tests were made with a 3,000 pound Universal testing maohine, but this prooedure was discontinued due to pressure of work for that machine. The hydraulio apparatus (calibrated against the Universal testing maohine) has given re-produoible results. It has the added advantages of being simple, inexpensive, and much faster i n operation. Three knife angles were compared} 30°, 60°, and 90°. The 90° angle was chosen for further work, although i t s superiority was not estab-lished conclusively. The knives were oarefully ground, then finished on an - 33 -o i l stone. Later work on stress analysis proved this ohoiee to be a con-venient one* It was found neoessary to grease the knife-edge before testing each specimen to keep pitch from collecting on the knife and altering the fr i o t i o n a l component of the applied force* The standard chosen for speoimen dimensions was 1*00" x 1.00" x 0.400", the latter measurement being the thickness of "plywood" obtained when two 0.200" veneers are glued together* This two-ply construction has been adopted as standard for oertain research work but for other purposes suoh as quality control testing i t has been necessary to use three or more ply oonstruotions* 3. EXPLORATORY TRIAL The Glueline-Cleavage Method was invented to yield more reproduc-ible results than those mechanical methods with which dissatisfaction has been expressed. In order to test the v a l i d i t y of this assumption an exper-iment was set up to compare the results obtained from three test methodst Tension Shear, Tension Normal, and Glueline-Cleavage. This comparison was oonduoted i n conjunction with a study of the effect of drying temperature on the strength of phenolic resin adhesive bonds with Douglas f i r veneers. Glue blanks 4 W x 4" x 0.200" thick were prepared with rotary out veneer from each of 26 Douglas f i r trees with a wide range of density and rings per inch. Twelve pairs of these side-matched glue blanks were prepared from each tree. Corresponding pairs from eaoh of the twenty-six trees were dried using a different sohedule for each of the twelve sets. After having been dried, a l l speoimens were allowed to reaoh equilibrium moisture content i n an atmosphere of approximately 50 per oent relative humidity and 70 degrees Fahrenheit i n preparation for bonding. Complementary pairs of veneers were then glued with the grain of the plies - 34 -parallel* Care was taken to ensure that a l l speoimens reoeived ldentioal treatment* One Tension Shear, one Tension Normal, and one Glueline-Cleav-age speoimen was out and tested from eaoh of these 4" x 4" laminates* This -work yielded twelve populations of laminates, a l l members of eaoh popula-tion having been treated ldentioal ly and then tested by eaoh method. Table 2 , Appendix A* presents the results of these tests expressed as Coefficients of Variation, These are considered to be the most reliable bases, of comparison* With this system the minimum Coefficient of Variation indioates the most reliable method of test* The Glueline-Cleavage Test yielded the lowest Coefficient of Variation i n every case* This evidenoe, based on 936 tests of speoimens matohed as nearly as i t i s possible to matoh wood speoimens, indicated that the Glueline-Cleavage Test was sup-erior and worthy of further intensive investigation. At this time i t appeared that the method had made use of the following principles which have been set down as prerequisite to an ideal tests 1, the test results are recorded i n pounds, whioh i s a widely aooeptable unit of measure, 2* the speoimen i s of the simplest possible shape, and therefore economical to prepare, 3. the load i s applied i n a manner such that the maximum stress i s developed i n the portion of the glue line just below the knife edge, 4. the equipment required i s very inexpensive, 5. the method of test i s readily adaptable for production testing as well as for researoh purposes, 6. the maximum possible number of speoimens oan be prepared from a given sheet of plywood. This desirable feature i s - 35 -made possible by the entire speoimen entering into the test. No material is wasted for gripping purposes. 4. EGxlO0 SPECIMEN^/ The exploratory tests proved so encouraging that further atten-tion was turned toward the design of a test specimen for research purposes. The f i r s t step when commencing suoh a design was to cheok theories regard-ing the requirements of the ideal method of test and test speoimen. Some of these could not be readily verified; e.g., the one regarding more uniform bond strength being obtainable with wood cells out at a slight angle. One theory whioh did lend i t s e l f to experimental confirmation was that one souroe of the differences between test results i s the varying per-centage of springwood and summerwood i n the glue lines of individual speci-mens. Tests were conducted on Douglas f i r specimens prepared with spring-wood bonded to springwood and summerwood to summerwood to verify this theory (see Tables IA, IB, and IC, Appendix A). It should be noted that the very • : An EGxlO 0 Speoimen i s defined as one which incorporates the following principles i n i t s design and manufacturej 1. It i s prepared from sawed and planed edge-grain veneers i n whioh the cells of the wood intersect the surface at an angle of 10° (see Figure 2, Appendix A). 2. The veneers are cross-banded, either at right angles or at some smaller angle, to provide for a more uniform distribution of springwood to springwood, summerwood to summerwood, and springwood to summerwood bonding than i s otherwise obtainable. For illustrations of 90° and 10° oross-banded veneers see Designs 1 and 3 of Figure 2, Appendix B. 3. The specimens are marked and sawn i n suoh a way that when the oleaving foroe i s applied by the knife the stress w i l l be concentrated i n that portion of the glue line immediately below the knife edge. This stress concentration Is aohieved, apart from the use of the knife edge, by a com-bination of the angle at which the wood cells interseot the glue line and the fact that wood i s relatively weak i n tension perpendicular to the grain. - 36 -limited numbers of speoimens were prepared from only one tree, consequently results should not be interpreted at this stage as indicating other than a probable trend. From the nature of the failures i n Tension Shear speoimens i t appeared logical to conclude that, i n springwood to springwood bonding, the wood broke i n tension at the base of the saw-out. This i n i t i a l tension failure was followed by a peeling or clearing action resulting i n complete rupture. In those speoimens bonded summerwood to summerwood, the wood was sufficiently strong i n tension to withstand failur e , and most of the bonds appeared to have been tested i n (tension) shear. The appearance of the broken Block Shear specimens indioated that the springwood to springwood specimens fai l e d i n compression para l l e l to the grain combined with shear through the springwood, whereas the summerwood to summerwood speoimens tested the bond i n (compression) shear. It appeared, therefore, that the type of failure obtained with springwood bonded to springwood was different i n appearance from that obtained with summerwood glued to summerwood. This difference i n appearance was also reflected i n the induoed stresses, the mean breaking load of summerwood bonded to summerwood being 90 per oent greater than that of springwood bonded to springwood i n the case of Tension Shear Tests (Table IA, Appendix A). It was 42 per cent greater i n the case of Block Shear Tests (Table IB, Appendix A). A " t t t test for the s t a t i s t i c a l significance of the differences between the means indicates that these are both highly significant (beyond the P - .01 l e v e l ) . These large differences i n breaking loads of springwood bonded to springwood and summer-wood bonded to summerwood may explain, i n part, the great variability enoountered i n test results obtained by standard methods. - 37 -The Coefficient of Variation of the breaking loads of springwood bonded to springwood calculated from Tension Shear speoimens was 15.4 per cent, and that of summerwood bonded to summerwood 14.4 per oent. When wood i s glued i n praotioe, some of the area i s bonded springwood to springwood, some summerwood to summerwood, and the balance springwood to summerwood. If these two samples are combined into one composite sample i t approaches (though exaggerated) that of normal test specimens. The Coefficient of "Variation calculated for this composite sample i s 35.3 per oent. The Co-efficient of Variation has thus been inoreased from 14.4 and 15.4 per cent to 35.3 per cent. Data for both Tension Shear and Block Shear methods are summarized below. Test Method Coefficients of Variation, per cent!/ Sp. to Sp. Su. to Su. Sp. to Sp. • Su. to Su. Tension Shear 15.4 14.4 35.3 Block Shear 14.8 16.9 23.9 These results demonstrate mathematioally how i t i s possible for the pattern of springwood and summerwood i n a bond to influenoe the va r i a b i l i t y of test results and how cross-banded edge-grain veneers may be used to reduoe var-i a b i l i t y for research purposes where this i s essential. From the Tension Normal data presented i n Table 1C, Appendix A, and the nature of the breaks, i t was observed that only one specimen broke i n the v i c i n i t y of the (Douglas f i r to Douglas f i r ) bond. The remainder failed either through the springwood or through the two additional glue 7/ -* See Tables IA and IB, Appendix A» for the basio data. Sp. and Su. are used as abbreviations for springwood and summerwood. - 38 -lines required to prepare the speoimen for test, therefore the bond under examination was not tested. This was due to imperfections i n the method of test, workmanship, or weakness of the wood. Theories.concerning this behavior have been outlined previously. As might be expeoted, the d i f -ference between the means of the breaking loads of the two samples did not prove significant. These data i l l u s t r a t e how rupture at some undesired place i n the specimen can be attributed to failure to ensure that the max-imum stress i s developed i n the glue l i n e . Although the experimental work upon which the following two pieoes of information are based w i l l not be presented u n t i l later i t seems approp-riate to use portions of the data at this time for i l l u s t r a t i v e purposes. The f i r s t i s the effect of concentrating the stress i n the glue line by con-t r o l l i n g the angle at which the wood oells are oriented with respect to the parallel to the glue line, to 10.978 when fibres were oriented at an angle of 10° to the glue line and when the knife was applied i n suoh a direotion as to concentrate the stress i n the bonded area. (The larger the slope ratio the more sensitive the t e s t ) . The second item concerns the effect of greasing the knife i n reducing v a r i a b i l i t y . The Coefficient of Variation was reduoed from 13.4 to 9.6 per cent^/. general approaoh wa3 correct and that the v a r i a b i l i t y within test results could be reduoed by making use of the principles already outlined i n 8 / — See Table 5A, Appendix B, for data; see (7) for definition of Slope Ratio. glue l i n e . from 8.012 when fibres were A l l results of the various comparisons seemed to indicate that the See Table 4. Appendix A. - 39 -Chapter III. The EGxlO0 Speoimen design^/ therefore incorporated as many of these principles as possible. This one inch square speoimen i s the simplest possible shape to manufacture, the only tools required being those normally available i n the average woodworking shop. Aoouraoy requires, of course, that these be i n good meohanioal condition. Suoh a specimen shape makes i t possible to cut the maximum number of test pieces from a sheet of plywood because no gripping area i s required and the whole speoimen enters into the test. These are worthwhile features when i t i s neoessary to pre-pare large numbers of speoimens from matched veneers. Another feature i s that maximum stress i s oonoentrated i n the glue line immediately below the knife edge. Some evidence i s at hand that a stress analysis of this type of speoimen is possible (33). Use is made of the different strength prop-erties of wood i n i t s various directions, notably i t s relative weakness i n tension perpendicular to the grain, to concentrate the stress i n that portion of the glue line immediately below the knife edge. The va r i a b i l i t y intro-duced by different speoimens having widely different percentages of spring-wood (or summerwood) i n their bonds has been minimized by oross-banding edge-grain veneers. The bond uniformity has been enhanced by using planed instead of peeled, sliced, or sawed veneers. The governing principle that the many details of gluing research must be performed according to pre-arranged standards has been used throughout. Broadly these inolude every detail of wood selection and treatment, glue selection and gluing, machinery, speoimen size and moisture content. - 40 -5. COMPARISON OF GLUELINE-CLEAVAGE (EGxlO0) WITH TENSION SHEAR, TENSION NORMAL, AND BLOCK SHEAR TESTS (A) Experimental Procedure In order to study the effeot of the above speoimen design i n reducing the variability of test results, and i n order to more fu l l y examine the preliminary conclusions, a further investigation was under-taken. This was intended to determine whioh of Tension Shear, Tension Normal, Block Shear, or Glueline-Cleavage with EGxlO0 specimens was the most suitable for evaluating the quality of Douglas f i r veneer joints bonded with a hot-press phenolio resin adhesive. It was essential that a l l test specimens be olosely matched i n every respect to permit a direct comparison of results. A l l specimens therefore were out from a single clear, straight-grained pieoe of Douglas f i r heartwood 4" x 4 n x 8* with the annual rings par a l l e l to one face. The selected pieoe of wood was as nearly flawless as oould be determined by visual inspection. Boards one-half inoh thick by four inches wide were sawed from the 4 W x 4" x 8' timber according to the pattern of Figure 2, Appendix A. These boards were end-ooated and allowed to air-dry for two weeks. They were then dried for one week at 122° F., to seven per oent moisture content, followed by two days at 144° F., whioh reduoed their moisture oontent to five per cent. This drying schedule produoed check-free lumber of uniform moisture oontent. The boards were transferred to a humidity chamber and conditioned to equilibrium moisture oontent i n an atmosphere of 32 per oent relative humidity and 70°F. They were then planed to produce veneers 0.200" j 0.005" thick, and f i n a l l y they were sawed to glue blanks 3.50" x 3.50". The thickness of each blank was measured at the four corners with a micrometer to a tolerance of jO.001". - 41 -This information was used when matching glue blanks to obtain layups of as uniform thickness as possible* Blanks for the preparation of EGxlO0 Specimens were matched, numbered and marked i n readiness for gluing, using the veneers out at an angle of 10° to the wood c e l l s . From the balanoe of the veneers, blanks for the preparation of Tension Normal, Tension Shear, and Block Shear specimens were matched, numbered and marked preparatory to gluing. Both Tension Normal and Tension Shear blanks were cross-banded while Block Shear blanks were laminated* The adhesive consisted of 2000 grams of PF512i a phenolic resin manufactured by Monsanto (Canada) Limited, 400 grams walnut shell flour and 400 grams water* The walnut shell flour was added to the resin and the mixture stirred for ten minutes. Following this the water was added and stirring continued for five minutes at which time the adhesive was ready for use. The i n i t i a l temperature of the resin was 19*5° C, and the f i n a l temperature of the adhesive was 24*0° C. This was spread on one face at 39 (S s 4)i2/ pounds per 1000 square feet of glue line with a mechanical glue spreader. Five minutes open-assembly time and five minutes olosed-assembly time were allowed, and pressure of 200 pounds per square inch pressure was applied at a platen temperature of 280 (S a 7)22/ degrees Fahrenheit for ten minutes. The glued blanks were then removed from the press and hot-stacked i n a olosed wooden box for 18 hours. Following this they were returned to the humidity chamber for reoonditioning (32 per cent relative humidity and 70° F.). After reaching equilibrium moisture content, the glued blanks n S n i s the standard deviation of the population as estimated from a sample. - 42 -were out into test specimens which were then returned to the humidity chamber for f i n a l reconditioning. The test specimens were cut and numbered according to the plan of Figure 4, Appendix A. Specimens were tested by their respective methods, immediately upon removal from the humidity ohamber. The following information was reoorded for eaoh specimen tested: (a) breaking load, i n pounds, (b) wood failure, i n per cent, (o) weight of the speoimen at test, i n grams, (d) oven-dry weight of the speoimen, i n grams, and (e) volume of the oven-dry specimen, i n oubio centimeters. (B) Analysis of Test Results See Table 3, Appendix A, for the data. It was anticipated that there would be l i t t l e difference i n the specific gravity of individual speoimens since a l l had oome from a single timber of uniform quality. This proved to be the oase so i t was unnecessary to oull any speoimens, or to apply a correction for specifio gravity differences. Likewise, moisture contents of the speoimens were sufficiently uniform so that no correction was necessary. Analysis of the breaking loads yielded the data presented i n Table 4» Appendix A. The Glueline-Cleavage Test (EGxlO0 Speoimens), with the knife greased, gave the minimum Coefficient of Variation, 9 . 5 per cent. Greasing the knife had reduced the Coefficient of Variation from 13.7 to 9 . 5 per cent, a very worthwhile reduction emphasizing the importance of greasing the knifo before each test. In addition to yielding the least variation between test results, the Glueline-Cleavage Test with EGxlO0 oross-banded speoimen has another - 43 -advantage; the glue line i s opened for inspection even i n a case where the wood i s "weaker than the glue". This feature may be observed by oomparing the photographs of broken Glueline-Cleavage specimens with those of the Tension Shear, Tension Normal, and Block Shear methods (Figures 5 to 8 inolusive, Appendix A). 6. SUMMARY The Glueline-Cleavage Test oonsists of measuring the foroe re-quired to cleave a one inch square test specimen with a 90° knife (wedge) plaoed along the glue line of the specimen. Almost any machine capable of applying and measuring the necessary load could be adapted to this method. The small hydraulic machine illustrated i n Figure 1, Appendix A, b u i l t at nominal cost, was used i n a l l of the Glueline-Cleavage tests herein reported. The EGxlO0 specimen has been designed to minimize v a r i a b i l i t y i n test re-sults for research purposes. The advantages of the Glueline-Cleavage Method (as separate from the additional advantages of the EGxlO0 Specimen) w i l l be summarized f i r s t . Undoubtedly the most important advantage of the method i s that i t has been shown to give more reproducible results ( i . e . , to be more accurate) than the other methods against whioh i t was compared. Second, the results are reoorded i n pounds, a universally acceptable unit of measure. Third, the specimens are relatively easy to prepare, requiring only the tools normally available i n any woodworking shop. Fourth, a maximum number of specimens may be out from a unit area of plywood, whioh f a c i l i t a t e s the use of experi-mental designs requiring large numbers of matched specimens. F i f t h , the region of maximum stress, and there is only one, is more d e a r l y defined than i n most other test methods. This i s advantageous when inspecting specimens for the causes of low f a i l i n g loads. Sixth, the method lends i t s e l f readily to production testing or "trouble-shooting". Seventh, a l l of the glue lines i n a partioular plywood speoimen oan be tested, regard-less of the number. Eighth, only inexpensive equipment is required. The EGxlO0 Specimen has two additional advantages. First and foremost, the variation i n test results i s less than i n any other style of speoimen tested. Second, a l l glue lines are opened for inspection even though the adhesive i s "stronger than the wood". CHAPTER 7 - ELABORATION OF THE GLUELINE-CLEA7AGE METHOD 1. PREAMBLE The Glueline-Cleavage Test, when used i n conjunction with EGxlO 0 Speoimens oross-banded at 90°, has been demonstrated to be a superior method for estimating the adhesive bond quality of plywood. The speoimens upon whioh this statement i s based were prepared so that they would have as uniform bond strength as possible. The question arose as to whether the method would maintain i t s superiority with a wider range of adhesives and weathering treatments. Apart from this general problem there were three speoiflo investigations whioh i t was desired to make. The fallacy of attempting to measure the strength of strong glue with weak wood has been considered. It may happen that a researoh worker does not know the relative strengths of the wood and adhesive which he proposes to use, but without this information i t is quite possible to draw erroneous conclusions from an investigation. A system whereby this p i t f a l l may be avoided would be of value. The development of suoh a method i s the f i r s t of the three speoial investigations noted i n the previous paragraph. The second point requiring examination was whether or not the design of the EGxlO0 Speoimen used i n Chapter 17 was the best. How muoh improvement, i f any, oould be effected by orossing the springwood-summerwood bands at an angle less than 90 degrees? What would be the effect of using EGx0°12/ or FGxO 0^/ veneers instead of EGxlO0 veneers? These questions appeared to require answers. i^EGxOO i s an abbreviation for edge-grain veneer (with the wood cells parallel to the surfaoe). 12/ — 1 FGxO° i s the corresponding abbreviation for flat-grain veneers. - 46 -The third problem was to demonstrate mathematioally that the results of mechanical tests were more reliable as an estimate of bond quality than were those based upon per cent wood failure , 2. AUXILIARY TOOLS Consideration was given to methods of setting up, conducting, and analysing the results of an experiment designed to investigate the problems outlined above. If an adhesive i s suspeoted of providing bonds "stronger than the wood", and i f i t i s desirable to verify this suspicion, then i t appears that some method must be found for preparing a glue that i s "weaker than the wood". This should not ohange the glue's nature so drastioally that conclusions are invalidated. To achieve suoh a purpose i n this researoh i t was considered advisable to add a standard mixture of extenders (walnut-shell flour, caustio, and water) to different percentages of the phenolic resin. The range of mixtures varied from that required by the manufacturer's formula (27&6 resin solids) to one known to be def-i n i t e l y "weaker than the wood" (12§jS resin solids). Any resultant design, then, must include more than one adhesive strength. The comparison of several possible speoimen designs would include a satisfactory measure of their relative efficaoies. The objective of suoh a comparison would be to determine which of several test speoimen designs and methods of test would be the most sensitive to small changes i n glue line quality. The design for a measure of sensitivity*'/ i s based upon the following premisej for an adhesive, suoh as the phenolic resin PF512 ex-tended with walnut-shell flour and water, there must be some combination of resin and extender whioh w i l l be only just as strong i n cohesion and/or —- "Measures of sensitivity" for the purpose of this thesis w i l l be defined as the a b i l i t y of the "measure" to distinguish between small changes i n the strength (quality).of a glue bond. - 47 -adhesion as the wood fibres which i t is to bond. Successively greater extensions w i l l produce bonds whioh w i l l f a i l at progressively smaller loads. Smaller extensions, however, w i l l not produce bonds which f a i l at increasing loads because the failure w i l l always ocour i n the wood and not i n the adhesive. In other words, the bonds formed by successively smaller extensions w i l l a l l break at the same load, that determined by the.strength of the wood. The f i r s t method investigated as a measure of sensitivity was the degree of correlation existing between breaking loads and per cent resin solids. After careful consideration i t was o'oncluded that this degree of correlation i s not the best available measure of test method sensitivity. A correlation curve represents the relationship between the variables, i n this oase breaking load as recorded by the chosen method of test and per oent resin solids. This, however, i s only of indirect interest. The primary interest i s direoted toward a reliable indicator of the f i r s t sign of reduoed bond strength as the resin solids content of the adhesive i s reduoed. This i s estimated by the rate at whioh the breaking load f a l l s off with decreased resin solids. More broadly, i t i s desired to be able to say that the test results truly represent the strength of the adhesive bond just before i t ruptures. It appeared, therefore, that the following points were of interest* f i r s t l y , the reductions i n breaking load rather than the absolute breaking loads, seoondly, the transformation of these figures to percentages so that comparisons oould be made between methods with accuracy, thirdly, the rate at which breaking loads decreased with reduced resin solids (i.e., the steeper the slope the more sensitive the test, other things being equal), and fourthly, the aoouracy of slopes, (i . e . , the re-producibility of results as measured by the Standard Error of Estimate). - 48 -A s t a t i s t i c a l technique known as "Slope Ratio A s s a y s " w h i c h i s used by medical men to determine whioh of several drugs is most effective seemed to be applicable and has been used with modifications. In essenoe, the procedure is to plot treatment effectiveness over dosages (the analogy would be breaking load over per cent resin solids) using a transformation such that the data w i l l plot as a straight lin e . The medicine giving the largest value of the ratio b/Se = Slope of line i s declared Standard Error of Estimate to be the most effective for each increase of dosage (for this purpose the largest b/Se would be declared the most effeotive for each reduction i n per cent resin solids). A comparison of accuracy between meohanioal and wood failure methods posed a problem, the data of the former being recorded i n pound units, and that of the latter i n per oent. A l l data must be reduoed to a percentage basis before direct comparisons may be made. In the oase of the mechanical test results, this was percentage of the maximum breaking load whereas the wood failure results are already expressed as per cent. If a valid correlation exists between bond strength and wood failure then per cent of maximum breaking load and per cent of wood failure should be directly comparable. It was decided to use this prinoiple i n analysing the data. The method used to compare the accuracy of bond strength predic-tions based upon mechanical tests and wood failure i s described below. I f specimens matched for bond strength are tested by two mechanical methods (e.g., Glueline-Cleavage and Tension Shear) at least four independent est-imates of bond quality may be made from them; two breaking loads and two •fowr per oent wood failures. A l l A o f these designs, i f they are to be useful, must record parallel trends when the bond quality i s altered. -* I f the strength of the bond has been reduoed by 25%, then each design should reoord - 49 -t h i s . The meohanioal tests should show a 25% reduction i n breaking load whereas the Per Cent Wood Failure Designs should show a 25% reduction i n per oent wood fa i l u r e . The difference (per oent reduction i n breaking load minus per oent reduction i n per cent wood failure) i s a measure of the incapacity of one (or both) to accurately portray this reduction i n bond strength. 3 . EXPERIMEMTAL DESIGN (STATISTICAL) The experiment was not laid out as a single s t a t i s t i c a l design. This would have been too complicated to be practical, but certain of the necessary principles were incorporated so that portions of the data oould be analysed by s t a t i s t i c a l procedures. For instance, care was taken that specimens from eaoh replication were included i n every treatment combin-ation. Nine designs of Glueline-Cleavage Specimen and four designs of Tension Shear Specimen were included. The details of these are illustrated i n Figure 2, Appendix B. Thirteen meohanioal estimates of the bond quality are provided, one from eaoh design. In addition, an independent estimate of the bond quality i s provided by eaoh design through the medium of the per oent wood failures read from the ruptured specimens. A total of twenty-six estimates of bond strength therefore were made available by this experiment. Not a l l of these have been used i n this thesis. Four strengths of phenolio resin adhesive, ranging from "stronger than the wood" to "weaker than the wood" were inoluded. These have been designated as Adhesives A, B, C, and D containing, respectively, 27g-, 22^, 17^, and 12-| per oent resin solids. It was decided to test speoimens after each of six increasing severities of the "aooelerated weathering oyole" (alternate boiling i n water and drying). Eight tree replications were included, one - 50 -specimen from eaoh tree being allotted to each of the treatment combina-tions. 4. WORKING PLAN Having f i r s t settled upon the treatment combinations the next step was to prepare a working plan i n d e t a i l . Materials were selected and the details of wood preparation inoluding sawing, seasoning, planing, and conditioning were performed, with one exception, aeoording to the standards set forth i n Chapter IV. Cutting to the pattern of Figure 1, Appendix B, instead of Figure 2, Appendix A, was the single exception. From the 0.200" veneers prepared with the fibres at an angle of 10° to the surface, twenty-four glue blanks 3.80" x 3.80" were sawed, to give eight for eaoh of the three Glueline-Cleavage designs using this material. Using the edge-grain veneers prepared with the fibres p a r a l l e l to the surfaoes, 64 glue blanks 3.80" x 3.80" were sawed, 24 for the three Glueline-Cleavage designs, 16 for the Tension Shear, and 24 for the Tension Shear Plywood designs. From the flat-grain veneers prepared with the fibres parallel to the surfaoes, 64 glue blanks 3*80" x 3.80" were sawed, 24 for the three Glueline-Cleavage de-signs, 16 for the Tension Shear, and 24 for the Tension Shear Plywood designs. The glue blank thicknesses were measured and reoorded on each oorner of the specimens, to *0.001". Glue blanks were matched on a thick-ness basis, for each speoimen design, so as to obtain "layups" with as nearly as possible the same thiokness. Each layup was marked distinctly to f a c i l i t a t e matohing of the speoimens with a minimum of confusion when gluing. For uniformity, a l l specimens were marked i n the upper l e f t corner, as shown on Figure 3, Appendix B. The number, representing the specimen design, glue mix, and tree, was placed on eaoh veneer of the layup. Specimen designs and names are ill u s t r a t e d on Figure 2, - 51 -Appendix B. Due to the volume of work, four days were required to complete the gluing. Because there were four grades of glue i t was convenient to al l o t one day to each glue type (glues and days were randomized). For the f i r s t two tress tested, the order of gluing the layups was chosen by a random sampling procedure the results of which are set forth i n Table 1, Appendix B. Replications were treated i n a similar manner. The four formulae below provided sufficient adhesive to glue a l l of the blanks whioh were bonded with that particular mix (i . e . , for eaoh day's gluing). In addition each provided sufficient for waste, as i t was neoessary to clean the r o l l s after every second or third press load. Forty pounds of adhesive per 1000 square feet of single glue line (1.82 grams-per 3.80" x 3.80" speoimen) was applied by a power-operated glue spreader. ADHESIVE "A"; 27^5 resin solids. This, the standard mix, contained 65.0# PF512 liquid resin or 27.6% resin solids. To 32 grams of oaustio soda dissolved i n 40 grams of water were added 520 grams of water at 200°F. These were stirred for one minute after which sti r r i n g was continued while 320 grams of walnut-3hell flour and 80 grams of soda ash were added. The mixture was then stirred continuously for 20 minutes, keeping the temperature above 180°F. Agit-ation was oontinued u n t i l the temperature was reduced to 140°F. Finally 2000 grams of PF512 resin, 2 grams of stove o i l , and 80 grams of walnut-shell flour were added and s t i r r i n g continued for 20 minutes during whioh time the temperature was further reduced to 70°F. The viscosity was then measured with a McMiohal Visoosimeter, and the adhesive was ready for use. - 52 -ADHESIVE "B"; 2 2 ^ resin solids This formula contained 22.5% resin solids. To 46 grams of oaustio soda dissolved i n 58 grams of water were added 759 grams of water at 200°F. These were stirred for one minute after whioh stirring was con-tinued while 468 grams of walnut-shell flour and 117 grams of soda ash were added. The mixture was then stirred continuously for 20 minutes, keeping the temperature above 180°F. Stirring was oontinued u n t i l the temperature was reduoed to 140°F. Finally, 1626 grams of PF512 resin and 2 grams of stove o i l were added and stir r i n g oontinued for 20 minutes during which time the temperature was further reduoed to 70°F. During the last 5 min-utes' st i r r i n g , the viscosity was adjusted (by the addition of a small quantity of water, or walnut-shell flour) to that of the standard mix, i.e . , Adhesive "A". The adhesive was then ready for use. ADHESIVE "Ctt; 1 7 ^ resin solids The procedure followed i n preparing Adhesive "Cn was the same as that for Adhesive "B" with the exoeption that the proportions of ingred-ients were adjusted so that the resulting mixture contained 17§?S resin solids. ADHESIVE "Dn; 12^S resin solids The prooedure followed i n preparing Adhesive MD n was the same as that for Adhesive "B" with the exoeption that the proportions of ingred-ients were adjusted so that the resulting mixture contained 12-^ 2 resin solids. Five minutes open-assembly time was allowed, followed by five minutes closed-assembly time. Pressure was applied for ten minutes at 285°F. and 200 p . s . i . The glued blanks were allowed to oool and then were conditioned for 96 hours i n an atmosphere of 32 per oent relative humidity - 53 -and at a temperature of 70°F. Figure 4 , Appendix B, shows the marking and cutting plans for the different designs. With EGxlO 0 Glueline-Cleavage specimens i t was essential that the direction of knife application be marked on the speci-mens before they were out. An effort was made to saw Tension Shear specimens i n such a manner that the lathe checks "pulled closed" during testing^ 6). One test specimen from each tree, and from every design and ad-hesive, was tested after eaoh of the following "aocelerated weathering" treatments} I. dry} I I . after 4 hours boiling i n water; III. after 4 hours boiling, 20 hours drying at 145°F. and 4 hours boiling, or, abbreviating, after 4,20,4; IV. after 4,20,4; 20,4; V. after 4,20,4; 20,4; 20,4; 20,4; VI. after 4,20,4; 20,4; 20,4; 20,4; 20,4; 20,4. With the exception of weathering treatment I above, a l l speoimens were tested wet after having been cooled to room temperature. For each specimen tested the following data, with the exceptions noted, were reoordedi (a) breaking load, pounds, (b) wood failure, per cent, (o) weight of the specimen at test, grams, (d) oven-dry weight of the speoimen, grams, and (e) volume, of specimens at test (by micrometer measurements), oubio oentimeters. Specifio gravity measurements obtained from speoimens whioh have been boiled i n water are of doubtful accuracy. The specific gravities of those specimens tested dry were, therefore, aooepted as being representative of the specific gravity of the glue blank from whioh each was cut. The same procedure was applied as a basis for calculating the percentage moisture content. In order to estimate the specifio gravity and percentage moisture content of eaoh glue blank, the specimen chosen for testing dry from that blank was weighed, i t s dimensions measured (*0.001n) immediately before testing and i t s volume oomputed. After having been tested, the pieces were oven-dried and re-weighed. The individual speoimen for any given test was chosen at random from among those cut from that particular glue blank. Those specimens from each glue blank which were to be boiled were kept together. This f a c i l i t a t e d the seleotion of one speoimen for testing from each glue blank after eaoh treatment. The broken speoimens were allowed to s i t faoe up on a table i n the laboratory for approximately 24 hours before the per oent wood failure estimation was made. The pieces of each test specimen and eaoh glue blank were f i l e d together, as were the trimmings, so that this material could be used for further study. The glu-ing and testing prooedure was repeated three times as outlined. Altogether eight trees were tested, two i n the f i r s t series and three with each rep-e t i t i o n . Each design of test specimen yielded two independent estimates of the quality of that glue l i n e . The f i r s t was the breaking load and the second the per cent wood failu r e . In this thesis the terms Design 1, Design 10, Design 13, etc., refer to breaking loads, percentages of max-imum breaking loads, or percentage reduction i n breaking loads. When estimates based upon per oent wood failure are being considered they w i l l - 55 -be referred to as Design 1$WF, Design 10$WF, Design 13?2WF, etc. Estimates of per oent wood failure are based upon the standard used by the Douglas F i r Plywood Association. 5. RESULTS The f i r s t step i n analysing the data was reduction of a l l f i g -ures to a common basis of comparison. The one chosen was the peroentage reduction i n breaking load (or per oent wood failure) whioh had been induced by each combination of treatment for eaoh tree and eaoh design tested. The reasons for this ohoice of transformation have been developed i n the section of this chapter entitled "Auxiliary Tools". In Table 2, Appendix B, i s presented the original data and i n Table 3, Appendix B, these data transformed to percentage reductions i n breaking loads (or per cent wood f a i l u r e ) . (a) Per Cent Wood Failure Vs. Meohanical Test For the purpose of this comparison wood failures obtained from Tension Shear (Plywood Shear) specimens w i l l be used. This i s a widely employed test procedure for wood failure estimates, and the system for estimating per cent wood failure was developed for this particular speoi-men, wood, and glue. Thus conditions are optimum for the use of per cent a wood failure as an estimate of bond quality. Data from Designs 4, 11, and 1 1 0 F w i l l be used for i l l u s t r a t i v e purposes. Design 11 vs. Design 11$WF Table 4A» Appendix B, l i s t the differences (per cent reduotion) (per cent reduction ) (in breaking load ) minus (in per cent wood failure) (of Design 11 ) (of Design ll^WF ) for each treatment combination and for eaoh tree tested. The mean of the differences is 5*5%, i.e., using the averages of 196 specimens per design - 56 -i t appears that the two designs gave approximately the same results. Design lljSWF indicated 5.5% greater reduction i n bond strength than did Design 11. Upon closer inspection of the data i t is found that the Standard Deviation of these differences i s j37.4$. Although the estimate based upon the mean difference appears reasonably accurate, any individual difference i s very unreliable (68 out of 100 of the differences being equal to or smaller than +37*4% and, of course, the other 32 out of 100 being larger than £.37.4$) • Design 11 vs. Design 4 By comparison, when the same technique i s used with the d i f -ferences (per cent reduotion) (per oent reduction) (in breaking load ) minus (in breaking load ) (of Design 11 ) (of Design 4 ) for matched speoimens (matched for treatments and tree) the mean of the differences i s 6.5%. The Standard Deviation of these differences i s •15.4$. See Table 4B., Appendix B, for the data. The average difference i n the estimates has changed from 5.5% to 6.5%. This differenoe i s probably not significant. The Standard Dev-iation, however, has been reduced from +37*4% to ±15»4-%t a very notable reduotion. Design 11 and Design 4 agree much more olosely than do Design 11 and Design ll^WF i n spite of the fact that the latter were made from the same specimen and the former ware made from different specimens. Suppose that two specimens were prepared from the same material (wood and glue), treated identically by a treatment calculated to reduce the bond strength, and then one tested by Design 11 and the other by Design 4. The expectation would be that the difference between the two meohanioal estimates (of the percentage reduction i n bond strength) would not exceed ±15.5% with - 57 -a probability of .63 or •30.8# with a probability of .95. Had two est-imates been made using the single speoimen tested by Design 11, one using Design 11 and the other Design 11#WF, the expectation i s that the d i f -farenoe (between the two estimates of the percentage reduction i n bond strength) would now not exoeed *37.4^ with a probability of .68 or +74.8^ with a probability of .95. The inescapable conclusion i s that per oent wood failure (Design ll^WF) i s not a reliable estimate of the bond quality of any single speci-men (although i t may have merit when an average i s taken of a great many specimens). The d i f f i c u l t y i s that only one speoimen i s examined at a time and the researcher attempts, oonsoiously or otherwise, to say that the estimate of per oent wood failure measures the quality of the bond i n that particular speoimen. Then, too, there i s an ever present pressure to reduoe the number of specimens for economic and praotioal reasons. This oould lead to incorrect conclusions, wasted researoh, or faulty' production oontrol. Per cent wood failure should, for researoh purposes, be used only i n conjunction with mechanical tests, not as a substitute for them. Should i t be necessary to use per oent wood failure as an estimate of bond quality this should be done only after the required correlation with meoh-anioal methods has been established, (b) Sensitivity of Designs A comparison w i l l next be made of the sensitivity of individual designs, one of the major objectives of this research being to establish whioh of several designs of test speoimen i s the most sensitive. This comparison w i l l be restricted to meohanical methods for reasons whioh have been established. When the analysis was commended the Slope Ratio Assays method - 58 -was applied to only part of the data, namely to Adhesives A and B (27i# and 22^> resin solids)* These data provided linear correlation without transformation and were i n that portion of the curve of most interest, namely, the region where a loss of strength was f i r s t notioeable* This approach appeared to he favorable; after further consideration, however, i t was deoided that a l l of the data (Adhesives A, B, C, and X>) should be inoluded* Transformation of the data was necessary to meet the linear correlation requirement of Slope Ratio Assays* The transformation ohosen w i l l be explained later* This deoision to use a l l of the data was prompted by an irregularity i n the data, namely, that Tension Shear results indicat-ed a tendency for the breaking loads of Adhesive A to be less than those of Adhesive B* This was oontrary to expectation based on preliminary tests, and contradictory to the results given by most, but not a l l , of the Glue-line-Cleavage tests* Furthermore, i t was not true i n a l l oases of Tension Shear Tests* It should be noted that although the trend was a3 desorlbed above, the differences between the means was i n no case significant* It i s possible, therefore, that this apparent trend i s a result of sampling var-iations and oould be mathematically discarded on that basis* This question of whether i t was lik e l y that Adhesive B (22j^S resin solids) would give greater bond strength than Adhesive A (2lj0t resin solids), was disoussed with the ohemists of the resin manufacturer* They reported that an adhesive formula containing 2 2 ^ resin solids had been subjected to faotory tests and they oonoluded that " i t w i l l not quite come up to weatherproof standards"• The inference i s that this adhesive did not produoe as good a bond as the standard mix. Adhesive A* There i s no apparent reason why the glue lines (of Tension Shear and Glueline-Cleavage Speoimens) prepared from the same sheet of plywood - 59 -should not have identical strengths* This being the oase their breaking loads should follow the same pattern* There i s no logical explanation why Tension Shear and Glueline-Cleavage breaking loads should show opposite trends i n this o r i t i o a l region* It was decided to proceed therefore on the assumption that there i s an aotual decrease i n the strength of the bond when resin solids are reduced from 27^S to 2 2 ^ and that the differences noted are the result of sampling "errors"• A logarithmic transformation of the per cent resin solids soale was chosen to allow the use of linear correlation analysis as required for Slope Ratio Assays* Figure 5, Appendix B, illustrates typical plots of individual curves* It may be observed that a straight line plot i s a compromise between what should be a slightly oonvex downward curve for the least severe accelerated weathering treatment, to a straight lins for the medium treatment, and a sli g h t l y oonvex upward' curve for the most severe aooelerated weathering treatment* Description of the Method Used In order to obtain the greatest accuracy when comparing test speoimen designs, comparisons should be made within eaoh treatment combin-ation* In order to present a visual picture of the data, percentage reduc-tions i n breaking load were plotted over per oent resin solids for eaoh of Designs 1 to 1 3 , and for eaoh accelerated weathering treatment* In spite of the faot that a logarithmic soale was used for per cent resin solids i t was decided to plot the data on a natural scale (see Figures 6 , Appendix B for the data of speoimens tested after accelerated weathering treatment I ) . This necessitated plotting the (straight) regression line as a curve* Examination w i l l show that the use of logarithmic transformation for the linear correlation analysis and subsequent plotting of the data on natural - 60 -co-ordinates serves two useful purposest (1) i t meets the linear correla-tion demands of Slope Ratio Assays, and (2) i t gives a true picture of the shape of the f i t t e d curve and the spread of the data about i t . Linear regression analyses were made between peroentage reduction i n breaking load and the logarithm of per oent resin solids for each of the above. The regression lines were plotted on the oorrespending graphs of Figures 6, Appendix B. The slope ratio, b/Se s Slope , Standard error of estimate was next oaloulated for eaoh of the above mentioned regressions and this figure added to eaoh graph of Figures 6, Appendix B. The next step was to graph the families of ourves for each de-sign; these are Figures 7, Appendix B. Upon inspection of these ourve patterns for different designs i t i s noticeable that certain designs, notably 1, 2, 4» 5. 6 and 11 have produced f a i r l y regular patterns'subh.as oould be expected from a consideration of the probable effects of the treat-ments. Others, notably 8, 9, 10 and 12, have the symmetry of their pattern broken by certain curves crossing others. This i s interpreted as an i n -dication that results obtained from these designs are less reliable than those obtained from the other designs. Additional study has been limited therefore to those designs which produced regular and logical curve patterns. Designs 3, 7 and 13, having relatively minor distortion i n this respeot, have been included for further analysis. Analysis of Varianoe of Slope Ratios Next, the slope ratios were arranged i n the form of Table 5A, Appendix B, and an analysis of variance (Table 5B, Appendix B) was made to determine whether or not the differences between the means were significant. When this was found to be so, a further comparison was undertaken between individual means or groups of means (Single Degree Comparisons). Table 6, - 61 -Appendix B, l i s t s the Single Degree Comparisons of most interest* The oomparison between Designs 1 and 13 (means 10.978 and 9*173 from Table 5A. Appendix B) fa i l e d to show a significant difference (between the means). Because Designs 1, 2 and 3 have much smaller differences there i s no significant difference between their means. Study of the comparisons l i s t e d i n Table 6, Appendix B, prompts the following conclusions} (1) the three Glueline-Cleavage Designs u t i l i z i n g EGxlO 0 veneers are significantly more sensitive than both the Tension Shear Designs and those Glueline-Cleavage Designs using either edge-grain or flat-grain material (comparisons of Designs 1 • 2 • 3 vs. 4 • 5 .* 6, 1 vs. 11, 3 vs. 11, I vs. 13, 3 vs. 13, and 1 • 2 • 3 vs. 1 3 ) i ^ / , (2) oross-banded specimens prepared from either flat-grain or edge-grain veneers exhibit no significant difference between the means of Tension Shear or Glueline-Cleavage methods (comparison of Designs 4 • 7 vs. I I • 13), (3) the data f a i l to yield significant differences i n sensitivity between designs 1, 2 , and 3 , the three EGxlO0 Designs, at the accepted level of significance. Inspection of the Slope Ratios tabulated i n Table 5A, Appendix B, w i l l show that i n every oase the Slope Ratio for Design was greater than that for Design 2 . I t i s highly probable that a larger sample would indicate Design 3 to be significantly more sensitive than Design 2 . Comparison of the Slope Ratios of Design 3 with Design 1, on the other hand, w i l l show that i n one-third of the comparisons the Slope Ratios of Design 1 exceed those of Design. 3 whereas i n the remaining two-_ See Figure 2 , Appendix B, for speoimen designs. - 62 -thirds the opposite was true* It seems lik e l y that further sampling would substantiate the theory that no significant difference i n sensitivity exists between -these designs* Glueline-Cleavage Designs 1 and 3 (speoimens making use of EGxlO0 cross-banded veneers), have been shown to be the most sens-i t i v e of the twenty-six designs inoluded i n this study. There i s a possib-i l i t y that Design 2 should be inoluded on an equal basis* Design 1 would, therefore, be the ohoioe when (oross-banded) plywood adhesive joints are under study. Design 3 (or Design 2 when more applicable to the work i n hand) would be the ohoioe when laminated adhesive joints are under study* (o) Glue Strength vs* Wood Strength The fallacy of attempting to measure the strength of a strong adhesive through the medium of weak wood has been discussed, and ene of the objectives of this work was to find a method of avoiding t h i s . A simple method i s proposed for establishing whether a given adhesive i s stronger or weaker than the wood whioh i t i s to bond* The prinoiple i s to use several formulations of the adhesive ingredients instead of the single mixture recommended by the glue manufacturer. Some of these are purposely made "weaker than the wood". A few t r i a l s w i l l soon establish which formulae are "weaker than the wood" and whioh are not. The interpretation to be taken from such data depends to some extent upon the relative strengths of the wood and of the glue bond or the glue. When the glue i s "weaker than the wood", and treatments have been suoh as not to alter the strength of the wood, any reduction i n strength would be attributable to the effect of the treatment upon the adhesion or oohesion of the glue* Different conclusions may be required for hot-press phenolic resin adhesives, whioh are known to be very inert chemical sub-stanoes when completely polymerized* These adhesives, when used to bond - 65 -Douglas f i r aeoordlng to the manufacturer's Instructions, form bonds whioh are "stronger than the wood". Any reduction i n strength induoed by a treat-ment may now be subjeot to two or more interpretations. One i s that adhesion between the wood and the glue has decreased. Another i s that the strength of the wood i n the v i c i n i t y of the bond has been reduoed. Should i t be necessary to differentiate between these, a determination of whether the adhesive i s or i s not "weaker than the wood" may provide a key to the so l -ution. 6. SUMMARY Estimates of plywood bond quality based upon the breaking loads of meohanioal tests have been shown to be superior i n aoouraoy to those based upon per oent wood f a i l u r e . The importance of determining whether the adhesive i s "weaker than the wood" or "stronger than the wood" has been stressed and a method for making this determination indicated. When used for "trouble shooting", "production testing", or sim-i l a r work where the EGxlO 0 Speoimen oannot be employed, the indication i s that the Glueline-Cleavage and Tension Shear methods are of equal aoouraoy. The Glueline-Cleavage Method, when used with EGxlO 0 Specimens, designed for maximum aoouraoy, has been confirmed as the most acourate method subjected to test. - 64 -CHAPTER 71. CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE RESEARCH The purpose of testing wood-to-wood bonds in. general and plywood bonds i n particular has been reviewed. A brief summary has been made of the history of glue bond testing. The excellent researoh of early workers^ has been summarized. Finally, reasons have been given for a continuing need of methods to estimate bond quality. Since established tests f a i l to meet this need within the required limits of aoouraoy there i s a neoessity to develop more definitive methods. The scope of this researoh has been limited to hot-press phenolic resin bonds of Douglas f i r veneers. I n i t i a l l y adhesive researoh workers used mechanical tests to measure bond quality. More recently, estimates based upon the percentage of wood failure i n a ruptured bond have become popular, especially for glue specification or quality control purposes. The mechanical test has been shown to be basic, tests based upon estimates of per oent wood failure being of value only after a reliable correlation has been established between the two methods. D i f f i c u l t i e s of interpreting wood failure have been reviewed and the neoessity for a world-wide exchange of standardized per oent wood failure specimens suggested. The fallacy of attempting to measure the strength of strong glues through the medium of weak wood has been pointed out. Some of the deficiencies of existing meohanical tests have been re-viewed. The continuing need for meohanioal methods of measuring the strength of adhesive bonds has been established. In addition, i t has been shown that the defects of mechanical methods may be corrected more easily than those of other methods. Mention has been made of the f u t i l i t y of attempting to obtain uniform pure shear, tensile, or compressive stresses across glue line test Anderson, Browne, Brouse, Hopkins, Lee, McBain, and Robertson. - 65 -areas* The role of stress concentrations i n premature failure has been reviewed* The most useful stress distribution i s one i n which the maximum stress occurs i n a (single) known portion of the glue line* A l i s t has been included of the d i f f i c u l t i e s to be overcome i n any effort to measure the stresses i n a particular specimen* The hope of successfully using stress analyses to reduce v a r i a b i l i t y (in spite of the above-mentioned d i f f i c u l t i e s ) has been expressed* Methods of reducing the vari a b i l i t y between breaking loads of specimens have been considered. These began with the recommendations of the Adhesive Researoh Committee^ 4), proceeded to those of various other authorities, and f i n a l l y incorporated some of the author's views. The role of s t a t i s t i c a l techniques, controlled stress concentrations, and controlled lathe oheok orientation i n reducing v a r i a b i l i t y has been noted. Other tech-niques includei f i r s t l y , transforming breaking loads of glue joints to per-centages of wood strength (for comparative purposes); seoondly, the use of several adhesive qualities to avoid the dilemma created by attempting to measure the strength of strong glue with weak woodj and f i n a l l y , employing the rate of reduotion i n strength induced by a series of accelerated weathering cycles of increasing severity* Theories regarding a few specific sources of exoessive v a r i a b i l i t y i n test results have been formed. The premature failure of Tension Normal specimens due to stress concentrations at undesirable locations i n the speoi-men i s the f i r s t of these* Others include the effeot of sawout depth, and the varying pattern of springwood and summerwood i n introducing excessive va r i a b i l i t y into test data* Theories have been expressed regarding the essential requirements of the ideal mechanical test for glue bonds* The major requirements whioh - 66 -have been, set arei f i r s t , that test.results be reproduoible; second, that results be expressed i n a universally reoognized measure of force or stress; third, that specimens be of a design simple to prepare to a high standard of aoouraoy; fourth, that the maximum number of speoimens be obtainable per unit area of plywood; f i f t h , that the machinery required be inexpensive; and sixth, that the method be adaptable to both researoh and production testing. Methods of reducing-the v a r i a b i l i t y due to certain suspected sources have been studied. These sources include the differing springwood-summerwood pattern between speoimens (and within different areas of a single speoimen), the orientation of the wood cells to the surfaoe of the veneers, the roughness of veneer surfaces before gluing, and the necessity for stand-ardizing every detail of machine and specimen design. The above-mentioned study resulted i n the development of a new method of test which has been named the Glueline-Cleavage Test. The method i s to measure the force required to oleave or s p l i t a specimen by means of the wedge. The action i s similar to sp l i t t i n g wood with a wedge except that the knife i s placed along the glue line instead of on a simple (or solid) block of wood. Figures 1 and IA, Appendix A, i l l u s t r a t e the maohine with a specimen i n position for testing. For produotion testing of plywood, one-inoh square speoimens are cut from the sheets at an angle of 45° to the grain. For researoh purposes, on the other hand, where aoouraoy i s of the utmost importance, speoial specimens have been designed. In these speci-mens the wood has been out so that the cells intersect the surface of the veneers at an angle of 10°. The method makes use of this small angle, plus the relative weakness of the wood i n tension perpendicular to the grain, to concentrate the stress i n a small area of glue line immediately below the - 67 -knife edge* Care must be taken to see that the knife i s applied to the correct edge of the glue line, otherwise the rupture w i l l tend to occur along the grain of the wood instead of down the glue line* When this i s done every specimen splits through the adhesive or at the wood-glue inter-face* Where cross-banded edge-grain veneers are used i n test specimens the var i a b i l i t y i s reduced beoause of the more equal distribution of the spring-wood-summerwood within each specimen and between specimens* In the exploratory t r i a l , using lathe-out veneer samples, the Glueline-Cleavage Test gave more reproduoible results than the Tension Shear or Tension Normal methods. A second comparison was made, this time between the Tension Shear, Tension Normal, Blook Shear (modified specimen size), and the Glueline-Cleavage Test Methods* EGxlO 0 Specimens with veneers cross-banded at an angle of 90° were employed i a this t r i a l . The latter t r i a l was conducted with conditions designed to yield minimum vari a b i l i t y between the breaking loads of different specimens* Again the Glueline-Cleavage Test gave the highest aoouraoy* I A third t r i a l was conducted to determine whether the Glueline-Cleavage Method would maintain this superiority over a wider range of bond strengths and treatments, and to compare the accuracies of a larger number of specimen designs. For this purpose nine Glueline-Cleavage and four Tension Shear designs were included. "Treatments" included four strengths i - -of hot-press phenolic resin adhesive and six severities of accelerated weathering. Replication was provided by including eight trees i n the sample. One speoimen from each tree was subjeoted to every treatment combination prior to test. Eaoh test speoimen yielded two independent estimates of bond quality» the breaking load of the meohanioal test, and the percentage of wood failu r e . In effect twenty-six estimates of the various bond qualities - 68 -were available for comparison, two for each test specimen design* Although estimates of bond quality based upon wood failure had been ruled out prev-iously on theoretical grounds, i t was thought worthwhile to compare them mathematically with mechanical test results* Meohanioal estimates of bond strength proved superior i n accuracy to those based upon per oent wood failure* The Glueline-Cleavage and Tension Shear methods proved equally aoourate when used with specimens of the type available for "production-testing" or "trouble-shooting" purposes* The machine used for Glueline-Cleavage tests was very inexpensive by com-parison with that required for the Tension Shear tests. This oould overcame a restrictive factor i n promoting quality control of the gluing process i n plywood manufacture. A method has been demonstrated for insuring that the relative strengths of wood and glue are known, information whioh may lead to more acourate conclusions regarding the nature of the bond fai l u r e s . The superior accuracy of the Glueline-Cleavage Test, when used with cross-banded EGxlO 0 Specimens, was confirmed. Other worthwhile advantages of the method are* (1) the glue line of every specimen i s exposed for inspec-tion, a most desirable feature, (2) test speoimens are easy to manufacture and consequently relatively inexpensive, (3) less testing time i s required than with the other methods compared, and (4) a maximum number of specimens may be prepared from a given sheet of plywood. This latter feature i s of value when i t i s necessary to use experimental designs involving large numbers of matohed specimens* A l l the above-mentioned advantages of the Glueline-Cleavage Test, with one exoeption, confirmed the findings of earlier t r i a l s . ' This exception related to the accuracy of the method when used for production-type testing. The f i r s t comparison indicated the Glueline-Cleavage Test - 69 -to be of superior accuracy, whereas the third t r i a l indicated only equality i n this respect. The data and analyses of the f i r s t and third t r i a l s are not directly comparable, nevertheless i t was anticipated that they would indicate the same trend. Further experimental work seems to be indicated to resolve thiB difference. In spite of the improved acouracy demonstrated for the Glueline-Cleavage Test the v a r i a b i l i t y between the breaking loads of matched speci-mens i s s t i l l disconcertingly large and a few observations on further experimentation might be appropriate. Research, by i t s very nature, un-oovers new problems during i t s progress. One which appears worthy of further investigation i s the stress distribution i n a test specimen. Suf-f i c i e n t work(45) has been done to indicate that this problem may be solved by using a mathematical approach. A more practical and related problem i s to develop a method that makes use of present knowledge regarding the stress distribution to reduoe the v a r i a b i l i t y between specimen breaking loads. This i s amenable to solution since the region of maximum stress i s known. Bond strength i s frequently correlated with specific gravity of the pieces bonded. A correction based upon the specific gravity of the wood at the most highly stressed portion of -the bond appears to offer a means of reduc-ing v a r i a b i l i t y between the breaking loads of otherwise matohed specimens. Probably a similar correction could be determined for variations i n the angle at which the wood oells intersect the highly stressed portion of the glue lines i n different specimens. Another point deserving attention i s the conversion of breaking loads to a common basis so that comparisons may be made between specimens whioh are not matohed for veneer thickness, density, and species. The lack of such a system has been one of the limiting factors i n the use of meoh-anical test methods. For oertain researoh purposes i t i s quite feasible to use a speoimen of standard dimensions, for oertain other purposes, however, this may be impraotioal. Plywood-bond quality oontrol frequently requires that tests be made on speoimens with more than one veneer thick-ness. Conversion of a l l data to a oommon basis for oomparison would be desirable. The author has demonstrated, by means of tests not recorded here, that breaking loads are correlated with veneer thicknesses. Once this correlation has been worked out to the necessary degree of accuracy i t should be possible to oonvert the measured breaking load to that expeoted with veneer of the ohosen standard thiokness. Likewise, i t should be possible to establish a similar system to compare bond strengths between different specific gravities or species. The author i s convinoed that researoh to further improve the accuracy and the adaptability of the Glueline-Cleavage Test, along the lines indicated above, would be a worthwhile contribution to further development i n the f i e l d of adhesives and adhesion. 71 BIBLIOGRAPHY American Society for the Testing of Materials (A.S.T.M.) Standards, Philadelphia, A.S.T.M., 1953. American Society for the Testing of Materials, Symposium on Adhesives, Philadelphia, A.S.T.M., 1945. Bensend, D.W., and Preston, R.J., Some Causes of Variability i n the Results of Plywood Shear Tests, Madison, Wisconsin, U.S. Forest Products Laboratory Report No. R1615, 1946. Bergin, E.G., The Significance of Wood Failure i n Glued Joints, Canadian Woodworker, March, 1953, (reprint). Bergin, E.G., and W.E. Wakefield, A Comparison of Test Methods for the Evaluation of Cold-Press Urea-Formaldehyde Resin Glues, Ottawa, Canada, Forest Products Laboratories of Canada, 1946. Bethel, J.S., and J.B. Huffman, Influence of Lathe Cheok Orientation on Plywood Shear Test Results, Sohool of Forestry, North Carolina State University, 1950. Blis s , C.I., and D.W. Calhoun, An Outline of Biometry, New Haven, Conneotiout, Yale University, 1951, Mimeo. British Standards Institute Specification B.S.1203, Synthetic Adhesives for Plywood (Phenolic and Aminoplastio), 1945. Brouse, Don, Factors Affecting the Test Value of Casein Water-resistant Plywood, Purdue University Master's Thesis, 1927, (unpublished). Chelvarajan, B.K., Comparative Study of Indian and American Plywood Shear Test Standards, Indian Forester, Vol. 80, No. 1, 1954. Commercial Standard CS 35-49, Hardwood Plywood, U.S. Department of Commerce, National Bureau of Standards, 1949. Cousins, F.W., Determining Stress i n Glued Joints by Photo-elastio Method, Timber News, Vol. 58, No. 2127, January, 1950. Department of Scientific and Industrial Researoh, F i r s t Report of the Adhesives Research Committee, London, His Majesty's Stationery Office, 1922. Department of Scientific and Industrial Researoh, Second Report of the Adhesives Research Committee, London, His Majesty's Stationery Offioe, 1926. Department of Scientific and Industrial Researoh, Third and Final -Report of the Adhesives Researoh Committee, London, His Majesty's Stationery Offioe, 1932. - 72 -(16) Department of Soientifio and Industrial Research, Report of the Com-mittee on the Mechanical Testing of Timber, London, His Majesty's Stationery Office, 1934. (17) Elmendorf, A.» Basio Tests for Plywood, Forest Products Research Soo-iety Proceedings, Vol. 2, 1948. (18) Elmendorf, A., Methods of Testing Wood Adhesion, Wood and Wood Products, May, 1952. (19) Frooht, M.M., Photo E l a s t i c i t y , New York, N.Y., John Wiley and Sons, 1949, 2 volumes. (20) Hopkins, R.P., Evaluation of Resin Adhesives, Forest Produots Research Society Proceedings, Vol. 3, 1949. (21) Kline, G.M. and F.W. Reinhart, The Fundamentals of Adhesion, Paper Trade Journal, Vol. 129, No. 26, December, 1949. (22) Knight, R.A.G., Requirements and Properties of Adhesives for Wood, Forest Products Researoh Bulletin No. 20, London, His Majesty's Stationery Offioe, 1950. (23) Knight, R.A.G., Doman, L.S., and Newall, R.J., Durability Tests on Plywood Adhesives - Series,II, F i f t h Year's Analysis, Investigations into Glues and Gluing, Progress Report Sixty, Prinoes Risborough, England, Department of Scientific and Industrial Research, Forest Produots Laboratory, 1951. (24) Knight, R.A.G., Adhesives for Wood, New York, N.Y., Chemical Pub-lishing Co. I n e , 1952. (25) Knight, R.A.G., Durability and Performance of Adhesives for Wood, Plastics Progress, 1953, (reprint). (26) Knight, R.A.G., The Assessment of Bond Quality i n Glued Joints, Part 1, Analysis of Series II Plywood Experiments, Princes Risborough, England, Department of Soientifie and Industrial Research, Forest Products Research Laboratory, 1953. (27) Laoey, P.M.C. and H.A. Howe, The Testing of Glues by the Glueline Method, Prinoes Risborough, England, Department of Scientific and Industrial Research, Forest Produots Laboratory, 1949 and 1950. (28) Levon, M., Proposal for the Establishment of Rules for the Standard-ization of Test Methods and Test Pieoes for the Strength Testing of Plywood, Paper to the Timber Researoh Committee of the International Union of Forest Researoh Organizations, 1939, (N.V.M.). (29) Marra, A.A., Glue Line Doctor, Southern Lumberman, Vol. 183, No. 2290, September, 1951. (30) Marra, G.G., and J.W. Wilson, Preliminary Report on Test Method for Evaluating Gluability of Hardboard, Pullman, Washington, Division of Industrial Research, Washington Institute of Technology, Washington State College, 1952. - 73 Maxwell, J.W., Shear Strength of Glue Joints as Affected by Wood Sur-faces and Pressures, Syracuse University, Teohnical Publication, No, 64, New York State College of Forestry, 1944. Newall, R.J., Tests Connected with the Proposed Revision of B.S.1203, Progress Report 63, Investigations into Glues and Gluing, Princes Risborough, England, Department of Soientifio and Industrial Research, Forest Products Laboratory, 1953. Northoott, P.L., Analysis of the Stress Pattern Induces i n a Plate of Infinite Width by a 90° Wedge Acting upon a Matching Groove i n the Plate, M.E.561, Vancouver, Canada, Mechanical Engineering Department, University of B.C., 1953* (unpublished). Northoott, P.L., The Development of the Glueline-Cleavage Test, Journal of the Forest Products Research Society, Vol. II, No. 5, Deoember, 1952. Orth, Otto G., Jr., Radio-frequenoy Gluing—A Research Project, Pacifio Plastics, October, 1947, (N.V.M.). Perkins, N.S., Predicting Exterior Plywood Performances, Forest Products Research Society, Proceedings No. 4, 1950. Royal Canadian Air Force Specification C-22-2 (A specification for the certi f i c a t i o n of oertain types of adhesives), Ottawa, Department of National Defenoe for A i r , 1942, Rudkin, A.W., A Simple Method of Testing Glue Lines i n Tension, Reprint No. 102, Australia, Council for Scientific and Industrial Researoh, Division of Forest Products, 1947. Selbo, M.L., and W.Z. Olseh, Durability of Woodworking Glues i n Dif-ferent Types of Assembly Joints, Journal of the Forest Produots Researoh Society, Vol. I l l , No. 5, 1953. Truax, T.R., The Gluing of Wood, U.S. Dept. of Agr. Bui. 1500, 78 pp., Washington, D.C, U.S. Government Printing Offioe, 1929. Truax, T.R., F.L. Browne, and Don Brouse, Significance of Meohanioal Wood-joint Tests for the Selection of Woodworking Glue3, Madison, Wisconsin, U.S. Dept. Agr. Forest Products Laboratory, 1929. Truax, T.R., Development of Wood Adhesives and Gluing Teohnie, Trans-actions of the A.S.M.E., No. 54, February, 1932, United States Forest Produots Laboratory, Veneer and Plywood, U.S. Dept. Agr., Forest Products Laboratory, Madison, Wisconsin, 1919. United States Forest Produots Laboratory, Madison, Wisoonsin, Effect of Heat on the Properties and Serviceability of Wood, Experiments on Thin Wood Specimens, 1945. Wakefield, W.E., The Tension Normal to the Glue Line Plywood Test, Mimeo. No. 121, Ottawa, Canada, Forest Produots Laboratories of Canada, 1947. 74 -APPENDIX A .5 - 75 -TABLE IA, APPENDIX A Tension Shear Tests of Douglas Pir Springwood Bonded to Springwood and Summerwood to Summer-wood - Using RCAF Specification C -22-2 Summerwood to Summerwood Springwood to Springwood Mean Breaking Loads $ Standard Deviations! Coefficients of Variation: Breaking Wood Breaking Wood Load Failure* Load Failure Lbs. % Lbs. % 915 20 415 100 945 5 520 100 800 100 460 100 930 100 620 100 850 100 485 100 950 95 545 100 1125 20 390 100 1135 90 510 50 1175 70 570 90 960 100 645 100 565 100 988.5 lbs. 520 .4 lbs. 142.2 n 8 0 . 0 n 14.4 per oent 15.4 per cent Note - The Coefficient of Variation of a l l the above specimens combined i s 3 5 . 3 per cent. •Percentage of wood failure was estimated using the Douglas Fir Plywood Association standard. - 76 -TABLE IB, APPENDIX A Blook (Compression) Shear Tests of Douglas F i r Springwood Bonded to Springwood and Summerwood to Summerwood - Using a Modified Speoimen Size l w Square Summerwood to Summerwo od Springwood to Springwood Mean Breaking Loads» Standard Deviationsi Coefficients of Variationi Breaking Wood Breaking Wood Load Failure* Load Failure* Lbs. • % • Lbs. - % 3550 90 2540 100 2830 95 2330 15 3700 65 2710 30 3870 100 2820 95 3480 50 2670 95 3650 65; 2330 1Q0 2240 30 1710 10 4020 95 2160 100 4020 85 2810 70 3484.4 lbs. 2453.3 lbs. 589.0 362.6 n 16.9 per oent 14.8 per oent Note - The Coefficient of Variation of a l l the above speoimens combined i s 23.9 per cent. •Percentage of wood failure was estimated using the Douglas F i r Plywood Association standard. - 77 -TABLE 1C, APPENDIX A Tension Normal to the Glue Line Plywood Tests of Douglas F i r Springwood Bonded to Springwood and Summerwood to Summerwood Summerwood to Summerwood Springwood to Springwood Mean Breaking Loadst Standard Deviations! Coefficients of Variation: Breaking Wood Breaking Wood Load Failure* Load Failure* Lbs. *. Lbs. % 570 100 645 100 690 100 770 100 745 90 610 100 490 100 790 100 490 100 740 100 395 100 87© 100 590 100 635 100 710 100 525 100 585.0 lbs. 698.1 lbs. 123.3 n 113.1 n 21.1 per oent 16.2 per oent 1f Percentage of wood failure was estimated using the Douglas F i r Plywood Association standard. - 78 -TABLE 2, APPENDIX A Coefficients of Variation of Breaking Loads -Exploratory Comparison of the Glueline-Cleavage Test with Tension Shear and Tension Normal to the Glue Line Plywood Test Coefficients of Variation - per oent Sample No. Tension Tension Glueline-Normal Shear Cleavage 1 38.0 29.1 13.4 2 41.1 28.2 23.5 3 31.4 27.5 16.6 4 31.8 28.1 18.9 5 32.4 25.5 18.6 6 30.0 25.6 16.4 7 28.0 22.4 16.9 8 26.6 19.2 15.4 9 42.5 24.0 14.3 10 60.1 26.2 16.5 11 40.1 • 27.6 20.3 12 48.4 24.5 17.6 Means t 37.5# 25.6% 17.4?? Notei Eaoh sample (for eaoh method of test) oonsists of 26 speoimens. Total number of speoimens tested : 26 x 3 x 12 s 936. Glueline Cleavage EG x 10° Glueline Cleavage EG x 10° Tension Normal to the Glue line Tension Shear Blook Shear Specimens. 90° Knife - Speoimens. 90° Knife -Not Grea oed Greased Before Eaoh Test Breaking % % M.C. S.G. Breaking % % M.C. S.G. Breaking % % M.C. S.G. Breaking % % M.C. S.G. Breaking % % M.C. S.G. Load Wood at Oven Load Wood at Oven Load Wood at Oven Load Wood at Oven Load Wood at Oven (lbs.) Failure Test Dry (lbs.) Failure Test Dry (lbs.) Failure Test Dry (lbs.) Failure Test Dry (lbs.) Failure Test Dry 440 80 4.3 .70 348 100 6.2 .66 1873 90 6,e .74 209 5 5.9 .69 267 15 5.5 .69 505 90 5.5 .68 295 100 5.7 .66 2038 100 6.0 .70 209 5 6.4 .67 254 10 5.5 .66 387 100 5.5 .68 309 85 5.3 .64 ! 1593 100 5.1 .67 260 5 5.2 .69 254 20 5.7 .67 455 100 5.1 .68 345 95 5.6 .64 1127 95 5.8 .73 222 5 5.4 .66 267 15 5.1 .69 450 100 5.5 .67 302 100 5.7 .58 1242 100 6.2 .72 228 20 3.9 .66 246 10 5.1 .69 150 100 4.e .68 310 100 5.5 .69 1203 100 5.1 .74 267 20 3.4 .66 222 5 5.2 .70 425 90 5.1 .67 300 100 5.0 .70 2035 100 7.3 .69 306 15 3.6 .69 235 5 4.7 .71 385 95 4.9 .67 290 100 5.8 .66 ! 1995 95 6.4 .69 260 5 4.9 .70 235 5 5.5 .68 300 ioo j 5.0 .68 325 95 5.1 .69 1 1748 95 7.4 .69 241 5 4.8 .66 190 5 5.5 .67 270 95 " 4.5 .69 280 85 5.6 .68 ! 1962 100 5.6 .62 267 5 5.4 .64 215 5 4.9 .68 345 85 4.1 .65 298 90 5.4 .68 1648 100 5.1 .68 319 5 4.9 .64 246 20 5.5 .68 345 100 ' 6.7 .68 282 80 6.1 .69 1 1797 85 5.2 .60 196 5 6.5 .67 ! 280 5 6.1 .69 415 100 ' 5.1 .64 318 100 5.0 .70 \ 1648 85 4.6 .59 228 15 . 5.0 .69 241 10 5.1 .66 370 100 4.7 .64 332 100 5.1 .68 1 2015 95 5.4 .62 254 15 5.2 .65 22e 10 5.4 .66 285 100 4.9 .65 323 100 5.4 .66 ' 1775 90 5.0 .62 196 10 4.8 .65 241 15 5.9 .68 440 100 4.6 .66 295 100 5.5 .72 1915 95 5.2 .66 280 15 5.1 .69 380 100 4.7 .66 306 100 4.8 .67 1495 75 5.5 .60 274 0 6.7 .68 430 100 4.8 .65 307 90 5.6 .71 | ,1572 95 8.C .57 222 10 5.7 .68 414 100 4.7 .55 327 100 5.4 .68 1420 100 5.4 .69 209 0 5.1 .69 267 100 5.0 .64 312 100 5.7 .69 1981 85 5.1 .66 202 10 5.4 .70 100 100 7.2 .65 320 60 5.3 .70 1648 100 4.3 .68 202 10 6.6 .66 267 100 4.8 .64 392 100 4.e .68 1315 95 4.4 .70 248 5 6.4 .68 352 100 5.6 .65 420 100 5.5 .68 1863 80 7.5 .71 280 25 6.7 .69 ' 335 100 4.7 .63 425 95 4.5 .69 1360 95 6.5 .70 260 5 5.0 .69 | 264 100 5.4 .64 267 90 4.5 .67 1805 90 5.8 .69 293 5 5.2 .66 265 100 6.4 .64 282 100 6.1 .65 ' 1055 100 5.0 .69 254 5 6.0 .69 213 100 4.1 .64 295 85 7.0 .68 2125 85 5.2 .69 1 228 20 4.5 .70 389 100 \ 5.8 .66 280 50 4.9 .64 1985 75 8.5 .67 • 260 5 3.7 ..68 | 280 100 \ 4.5 .'65 310 80 6.0 .64 2515 85 3.9 .68 215 20 4.1 (.69 I 380 ioo ! 6.0 .65 215 95 6.1 .65 2215 95 5.1 .68 222 20 3.8 .66 337 100 4.6 .65 280 90 5.7 .68 1325 90 5.7 .66 260 10 5.1 .68 185 95 4.8 .63 305 60 5.0 .69 2105 95 6.2 .67 241 5 4.6 .69 365 95 5.6 .64 280 40 5.1 .69 1540 85 5.2 .68 267 5 5.0 .70 j 355 95 6.9 .66 233 70 5.2 .69 2185 95 6.5 .68 306 10 7.0 .68 245 100 4.6 .68 275 95 5.5 .69 2290 90 5.0 .68 254 10 5.3 .65 N= 35 35 35 35 15 x = 337.4 97.7 5.16 .655 308.1 89.4 5.45 .674 1754.6 92.4 5.74 .672 246.8 9.6 5.21 .676 241.6 10.3 5-38 .681 <T= 91.8 42.2 356.2 33.1 23.0 M.C. S.G. Moisture Content Speoifio Gravity TABLE 3, A - 80 -TABLE 4, APPENDIX A Statistics Mean No. of Mean Standard Coefficient Percentage Speoimens Breaking Deviation of of Wood Tested Load Variation Failure "N" (Pounds) (Pounds) (Per oent) Tension Normal to the Glue Line 97.7 35 337.4 91.8 27.2 Tension Shear 89.4 35 308.1 42.2 13.7 Block Shear 92.4 35 1754.6 356.2 20.3 Glueline-Cleavage EGxlO 0 Speoimens (knife not greased) 9.6 35 246.8 33.1 13.4 Glueline-Cleavage EGxlO 0 Specimens (knife greased) 10.3 15 241.6 23.0 9.6 --82 - _ F \"G. 2, APPENTMX A s SPRINGWOOB S U KIM ETC WOOD LUEUNE DIRECTION OF WOOD FIBRES S Q U A R E S SAWN l'* \" (FINISHED SIZE) f 1 . ARROWS, INDICATING DIRECTION IN WHICH KNIFE IS TO PENETRATE GLUELINE, TO BE MARKED ON EACH S Q U A R E BEFORE SAWING COMMENCES F O R G L U E U H F C L E A V A G E : % E G A 1 0 S P E C I M E N S J A N U A R Y 1951 F I G . 3 , A P P E N D ! * A T E N S I O N S H E A R T E N S I O N N O R M A L G L U E L I N E C L E A V A G E A N D B L O C K S H E A R P A T T E R N S F O R N U M B E R I N G A N D C U T T I N G T E S T S P E C I M E N S F I G . A, APPENDIX A 86 - 87 -TABLE 1, APPENDIX B Press Load No. Glue Blank Numbers First Day (Glue "AS) Second Day (Glue «DH> Third Day (Glue "CT") Fourth Day (Glue "B") 1 6A-1 1QA-1 10A-2 4A-2 2 U - l 1A-2 9A-2 12A-1 3 11A-2 13A-1R 1U-2R 13A-1 4 12A-2 2A-2 5A-1 8A-2 5 5A-2 6A-2 9A-1 7A-2 6 13A-2 11A-1R 13A-2R 11A-1 7 2A-1 3A-2 8A-1 3A-1 8 4A-1 7A-1 mm ™ 1 2D-1 4D-2 10D-1 5D-2 2 6D-1 3D-2 3D-1 2D-2 3 11D-1 11D-1R 13D-1R 11D-2R 4 1D-2 10D-2 5D-1 8D-1 5 6D-2 4D-1 12D-2 7D-2 6 12D-1 7D-1 8D-2 9D-1 7 13D-2R 11D-2 13D-2 13D-1 8 1D-1 9D-2 mm — 1 5D-2 12C-1 6C-1 3C-1 2 13D-2R 11C-1 13C-1 11C-2R 3 1C-2 7C-1 10C-2 3C-2 4 12C-2 8C-2 4C-2 10C-1 5 13C-2 11C-IE 13C-1R 11C-2 6 60-2 1C-1 8C-1 70-2 7 9C-2 5C-1 9C-1 4C-1 8 2C-1 2C-2 — — 1 3B-2 8B-2 1B-2 12B-1 2 11B-2R 11B-2 13B-1R 13B-2 3 10B-1 13-1 4B-2 7B-1 4 9B-1 6B-2 12B-2 2B-1 5 5B-1 10B-2 4B-1 8B-1 6 11B-1 13B-2R 13B-1 11B-1R 7 6B-1 3B-1 2B-2 5B-2 8 7B-2 9B-2 - -1 GLUELmE-CLEAVAGE B3SAKIK3 LOADS iVoattinring Adhnslve Cycle * 2 a 10 11 ' 1 235 210 260 345 II 135 155 195 175 A III 110 n o 195 175 IV 120 110 130 105 V 120 70 110 31C 71 20 45 45 20 I 215 230 255 315 II 150 125 145 \X> D 111 120 130 us 140 IV 45 125 ICO U 5 V 110 100 85 125 VI 10 65 80 125 I 210 195 250 285 II 30 50 90 20 C III 100 20 65 10 IV 15 0 90 10 V 35 0 3 0 VI 0 0 0 0 I 160 175 20 170 11 5 90 3 5 D III 12 0 0 0 IV 0 0 0 0 V 0 0 0 0 VI 0 0 0 0 12 13 14 15 2 a 10 11 295 270 235 230 540 610 440 400 165 155 150 190 255 3 » 275 415 155 145 120 125 215 215 215 230 105 155 100 150 240 290 165 185 l?5 125 no 125 125 230 195 105 105 100 100 165 165 K'5 185 155 260 240 230 175 -"15 215 215 190 195 235 215 220 515 610 430 450 390 U O 140 1/.0 165 190 290 30) 285 305 345 125 100 120 160 210 250 215 230 230 105 90 U 5 270 230 255 210 215 175 85 85 125 140 155 215 195 215 215 100 70 125 100 120 165 195 185 170 150 210 190 330 325 385 320 335 360 30 45 130 175 250 335 215 335 175 100 70 90 40 110 150 5 155 120 35 5 80 2 22 165 0 7 » 23 5 0 13 8 8 145 0 u o 0 0 0 0 31 5 30 0 10 0 175 125 165 15" 305 270 J05 270 295 240 255 15 a 8 10 11 12 13 L4 15 2 8 10 u 12 13 14 15 2 8 10 11 12 13 14 15 360 3B5 455 425 540 470 370 425 370 220 240 UO 260 240 210 195 230 380 360 410 u o 390 335 315 315 320 340 250 345 340 300 295 340 345 130 155 170 220 150 11D 70 175 345 250 315 335 300 285 270 355 305 210 250 250 270 255 255 270 260 155 105 120 165 195 105 120 L20 215 220 285 270 205 240 235 275 300 295 305 205 230 230 275 235 270 125 85 105 130 125 110 90 110 250 290 270 270 205 240 230 220 230 190 270 220 190 220 215 190 240 150 S 85 150 140 100 70 140 275 215 240 215 210 135 215 750 2fO 170 150 170 185 210 ie5 175 230 105 0 80 120 140 85 55 140 170 65 185 190 190 210 205 230 435 400 365 415 4*tt 435 360 400 3*5 215 175 300 295 260 205 190 150 340 410 455 500 415 340 355 285 340 360 365 365 300 315 280 295 335 145 155 185 175 UO 145 150 155 235 305 425 415 240 305 285 320 2B0 195 125 260 210 240 260 255 235 120 85 130 150 130 110 105 130 255 203 305 295 240 255 215 240 270 235 50 285 235 305 250 250 270 125 20 140 145 130 90 105 120 215 140 340 210 240 205 220 770 275 16s 150 280 190 215 210 150 205 110 140 125 120 35 IDG 130 240 200 255 235 310 190 215 255 230 130 140 190 175 175 175 190 205 0 0 110 125 110 30 65 105 175 5 220 185 175 175 215 270 325 290 300 400 390 380 285 340 345 215 195 295 285 275 170 155 185 365 430 440 430 510 340 325 335 270 125 200 285 250 240 185 240 220 120 140 155 45 170 110 60 155 240 190 340 305 345 270 320 305 175 20 25 150 100 185 150 140 205 90 40 n o 110 125 LOO 90 120 185 UO 305 240 250 215 185 150 85 0 0 165 45 15 35 165 150 LOO 5 65 90 105 45 80 85 170 0 215 140 205 65 140 120 110 0 0 125 10 20 0 125 65 110 0 18 , 0 30 0 45 35 215 0 210 15 205 0 130 55 10 0 0 10 0 0 0 0 20 15 0 0 0 10 0 20 40 15 0 20 20 20 10 70 30 185 270 230 300 235 195 250 235 230 175 28 210 230 220 175 H5 165 345 295 365 325 255 250 250 275 45 185 120 2 42 7 80 155 175 25 1 2 18 10 33 110 10 165 155 205 185 175 155 155 145 0 0 0 0 0 0 0 7 0 10 0 0 0 22 20 12 6 5 8 60 10 110 IB 140 8 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 10 0 0 25 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 CD CD A B Accelerated 6 Weathering Tree No. Cycle 2 8 10 11 12 13 14 I 380 300 410 550 410 340 305 II 280 360 300 385 280 255 280 III 195 260 240 260 iS 240 205 IV 235 190 240 215 <lr 235 220 V 215 305 235 195 -30 230 215 VI 230 195 165 190 220 1?5 205, 1 280 330 585 675 355 385 335 : i 360 335 325 365 295 250 275 111 200 200 340 280 210 250 215 IV 260 190 280 275 235 215 210 V 170 20 255 165 255 175 230 VI 215 2 210 175 175 170 210 1 365 345 405 415 345 435 295 11 385 350 340 380 365 240 220 111 140 175 195 260 215 185 205 n 88 210 185 140 205 105 170 V 35 10 55 60 120 20 80 VI 5 2 0 30 240 0 0 1 235 330 315 305 285 295 275 11 185 230 20 30 185 175 185 111 5 6 1 15 35 13 145 IV 0 0 0 . 0 25 0 5 V 0 C 0 a 0 0 0 VI 0 0 0 0 0 0 0 15 2 3 10 11 12 13 .14 15 2 8 10 280 195 215 255 320 305 145 175 230 385 350 365 320 165 145 170 170 195 45 170 175 300 270 400 205 135 165 175 135 165 100 125 125 320 260 270 215 110 110 175 175 U5 90 150 150 200 200 255 255 120 90 L40 US U.0 70 140 145 300 270 250 220 100 103 125 45 140 100 150 145 230 215 240 345 215 190 270 215 260 165 185 175 360 360 415 285 150 150 205 185 195 45 130 170 240 200 360 215 140 n o 145 190 145 120 125 110 260 210 315 220 120 70 130 150 130 110 n o n o 210 150 270 280 110 90 130 UO 105 100 125 130 275 175 270 215 145 15 20 130 120 70 125 110 200 145 250 300 185 125 280 300 275 170 190 165 350 325 520 300 125 85 140 3C 190 10 80 n o 195 175 320 175 105 80 120 105 125 90 100 110 200 130 260 155 30 20 105 120 120 55 80 100 190 185 280 120 90 0 90 40 45 05 125 100 190 165 205 45 125 0 10 0 10 70 35 100 70 5 220 •260 150 165 165 205 215 105 145 155 295 270 255 175 5 105 30 10 20 10 10 60 165 45 55 7 85 29 3 3 45 40 45 90 100 13 110 0 5 0 0 0 28 0 15 70 1 30 10 0 0 0 0 0 0 0 0 55 0 0 0 0 0 0 0 0 0 a 0 20 0 0 0 1 9 Tro. » tio. Troe No. 11 12 13 14 15 2 e 10 U 12 13 14 15 365 410 410 320 355 305 380 345 425 410 240 285 295 360 280 300 320 335 260 410 295 355 220 220 240 360 320 305 235 260 275 280 25i 365 385 250 175 185 250 295 260 280 240 240 250 270 280 340 175 255 240 315 260 250 260 305 280 270 370 270 275 230 185 215 355 260 250 280 280 335 295 105 295 300 210 190 230 305 490 385 355 300 365 350 320 365 390 385 255 325 315 320 320 240 255 345 320 185 260 315 250 305 270 305 285 275 240 250 285 230 170 255 240 275 205 230 240 295 260 235 175 250 235 210 290 360 305 230 230 230 240 240 250 215 275 240 190 190 210 260 235 230 305 150 280 275 205 305 215 150 215 220 250 190 130 280 320 390 300 190 280 345 280 435 335 435 295 325 315 295 335 235 235 275 295 275 305 260 230 175 210 285 205 270 215 175 205 200 145 295 250 305 215 195 210 155 190 185 140 195 210 175 295 365 255 165 145 210 80 195 210 140 210 230 110 195 85 130 175 UO 150 20 205 165 165 170 125 110 190 10 10 155 140 205 305 305 260 210 280 345 240 345 315 275 255 260 230 10 215 145 145 170 280 110 15 140 130 155 145 140 90 U.5 130 100 125 125 9 n o 15 70 100 130 120 25 110 85 45 125 5 0 70 0 15 55 120 UO 0 5 0 0 no 2 0 15 0 20 0 25 60 0 0 0 0 105 0 0 0 0 0 C 25 20 T A B L E 2 , A P P E N D I X B 10 Accelerated 'iJealharin^ j Tr« >• No. Adhesive Cycle 2 a 10 11 12 13 14 15 2 8 I 586 uo 693 429 6U 567 392 439 205 250 II 4.50 400 359 3B9 421 390 220 305 149 182 A III 350 332 234 365 382 310 2J5 232 205 >6 IV (.00 310 273 349 329 200 298 112 104 71 V UO 300 353 306 282 337 246 241 157 37 n 373 281 348 289 283 311 225 208 125 0 i 370 503 7ia 522 614 382 502 392 318 266 II 350 450 470 403 305 285 2V8 208 13-r 194 B i n 330 350 545 302 372 245 115 174 114 12b IV 320 330 251 396 324 303 292 248 145 4 V 313 313 462 513 277 438 245 260 74 0 VI 357 JOO 421 465 290 335 237 290 50 0 I 472 592 585 564 562 559 472 317 208 24e II 400 450 355 311 415 358 272 245 119 124 C III 350 217 373 302 304 304 295 220 161 0 IV 325 210 396 274 301 326 245 172 50 0 V 3W 200 301 330 313 345 244 272 0 0 VI 360 5 280 80 302 196 242 198 0 0 I 428 390 472 443 495 5U0 523 383 216 30 II 350 350 161 260 231 238 127 249 62 0 D III 250 100 142 0 515 76 155 271 0 0 IV 0 0 0 0 29 0 125 0 0 0 V 0 0 0 0 0 0 0 0 0 0 VI 0 0 0 0 0 0 0 0 0 0 10 Accelerated H'SJIUI wing True Ko. Adhesive Cycl« 2 8 10 11 12 U 14 15 2 8 I 100 ' 100 100 100 95 ICO ICO 100 100 100 II 100 ICO 1JO 100 ICO 100 100 100 100 90 A III ICO 100 100 100 100 100 ICO 100 10 100 IV ICO 100 100 100 100 1O0 100 100 0 95 V 100 ICO 100 ICO 100 ICO ICO 100 10 85 80 VI 100 100 100 100 100 100 100 100 10 I 110 100 ICO 100 ICO ICO ICO 100 85 50 II 100 100 100 100 100 100 100 ICO 0 100 fi i n 90 100 85 100 70 100 100 ICO 0 0 IV 50 100 ICO ICO 100 100 100 100 0 0 V 10 ICO 85 95 100 AO 100 100 0 5 VI 10 70 100 100 100 100 100 100 0 0 I 70 100 ICO 100 ICO 100 100 100 • 90 100 II 40 103 100 100 95 70 95 100 0 0 c III 10 as 100 ICO 100 95 50 100 0 0 IV 0 70 0 70 100 80 BO 100 0 0 V 0 60 70 0 ICO 25 100 100 0 0 VI 0 55 5 0 100 0 0 100 0 0 I 0 ICO 25 0 0 100 100 100 0 5 II 0 0 0 0 0 0 0 0 0 • D III 0 0 0 0 0 0 0 0 0 0 IV 0 0 0 0 0 0 0 0 0 0 V 0 0 0 0 0 0 0 0 0 0 VI 0 0 0 0 0 0 0 0 0 0 t o n : . 3 u .1 1/ , M i ' . -K.vt 11 12 Tr«« Mo. 1Y. >•* No. 10 11 12 13 14 15 2 8 10 11 12 13 14 304 236 240 177 254 159 U2 601 '635 505 498 481 4t4 229 U3 175 100 105 192 4J6 500 U8 411 4-.'-! 337 320 211 156 187 167 106 108 400 410 390 394 362 UO 322 102 152 147 uo 167 173 400 410 396 U6 325 354 353 148 183 112 115 83 155 400 410 430 385 227 379 261 107 72 82 83 84 123 367 400 483 404 410 358 357 336 a6 309 279 274 329 500 432 677 541 516 412 308 208 243 149 172 178 145 450 450 537 513 434 311 450 193 254 171 125 121 181 460 40J 498 435 437 370 376 140 165 205 123 123 110 450 350 577 492 431 372 320 13-! 133 132 62 104 158 503 286 467 502 337 248 414 58 106 93 65 52 120 450 410 427 512 425 342 348 339 - 291 219 224 133 241 480 683 729 534 482 543 4J2 214 219 126 155 110 136 450 500 435 314 413 492 286 85 148 174 39 122 76 400 404 340 442' 432 175 247 25 77 128 0 85 27 400 250 422 268 236 326 227 0 0 64 0 0 18 400 146 418 331 478 257 337 0 0 27 0 0 0 385 100 356 ICO 346 235 323 214 217 226 128 I06 186 440 553 577 494 533 369 370 0 0 33 18 18 8 400 300 275 236 322 130 218 0 0 0 0 0 0 377 196 143 118 216 84 136 0 0 0 0 0 0 ICO ICO 157 3 45 40 53 0 0 0 0 0 0 0 0 0 0 0 0 38 0 0 0 0 0 0 0 0 0 0 0 0 38 PER CffiT :«XD FAILURE 0? TEND ION ZHEAR TE3T Design No. 11 12 Tr e» No. Tr-e No. 10 11 12 13 14 15 2 6 10 11 12 13 14 60 100 100 100 80 100 ICO 100 100 100 ICO ICO lOO 100 UO 100 90 100 60 100 100 ICO ICO 95 ICO 100 35 100 HO ICO 100 100 ICO 100 ICO 100 95 60 100 ICO ICO 100 60 80 ICO 100 ICO ICO ICO ICO 95 90 40 101 90 100 20 100 100 ICO ICO 100 100 100 100 30 ICO 100 100 70 ICO ICO 50 85 100 100 90 75 100 ICO 100 100 95 10 100 100 90 95 100 ICO ICO 100 20 ICO 20 10 100 90 90 90 75 90 100 15 • 100 50 ICO 0 5 10 90 85 90 100 85 ICO 90 300 15 100 100 0 100 ao 60 85 80 ICO ICO 95 25 100 100 0 5 0 10 40 90 75 ICO 100 75 0 100 100 15 100 100 BO 100 100 60 100 95 85 100 100 100 20 100 25 100 100 65 100 100 ICO ICO 25 50 0 100 0 0 0 70 70 eo 0 95 5 0 0 0 0 0 0 15 50 95 0 90 0 30 0 0 10 0 0 0 0 30 10 0 100 U. 0 0 0 100 0 0 0 0 0 15 0 25 IC 15 0 0 0 0 0 0 0 5 10 0 • 90 0 0 0 0 50 0 0 0 60 95 0 0 100 5 100 0 0 0' 0 0 0 0 0 0 0 0 30 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 5 13 Tro< 1 Uo. 15 2 8 10 11 12 13 14 15 391 232 274 286 306' 231 230 220 234 335 224 295 222 209 202 147 220 226 302 214 228 228 237 176 212 205 192 404 i6e 164 156 227 165 130 213 215 369 180 212 166 158 172 215 150 156 353 140 78 137 226 235 219 203 lea 340 268 217 223 259 251 269 196 323 367 219 180 240 242 193 221 221 199 317 200 168 19S 241 215 173 121 203 J28 259 142 242 238 231 163 190 165 362 192 4 165 139 159 168 110 155 276 184 0 164 151 136 97 168 192 435 337 328 258 266 194 191 237 209 359 202 34 178 196 222 • 175 122 187 247 183 U 152 207 219 147 142 170 335 130 0 95 107 213 75 142 147 333 38 0 35 0 105 113 82 124 285 0 0 74 0 90 0 8 85 396 145 100 175 301 221 173 185 219 269 48 0 21 47 55 0 51 37 287 32 . 0 0 0 0 0 25 0 215 0 0 0 0 0 0 0 0 77 0 0 0 0 0 0 0 0 77 0 0 0 0 0 0 0 0 c c I 15 Tre. e Ho. 2 8 10 LI 12 13 14 15 100 100 100 ICO 95 100 100 ICO 100 ICO 0 50 95 eo 90 50 30 80 50 ICO 10 5 95 30 100 20 95 95 1(JU 5 10 95 ICO 20 65 20 95 25 5 10 75 ICO 50 85 100 100 20 0 5 25 75 50 to 30 ICO 10 100 100 100 ICO 100 10 90 100 ICO 50 5 15 95 10 60 10 90 0 100 60 10 80 5 50 20 100 0 60 85 0 5 70 10 60 100 0 c 35 50 100 50 95 10 5 100 0 0 5 0 0 30 00 100 0 95 100 100 100 100 100 100 PO 0 0 0 0 0 0 0 0 95 0 0 0 0 0 0 0 0 110 0 0 0 0 0 5 0 0 100 0 0 0 0 0 0 0 0 20 0 0 0 0 0 0 0 0 ICO 0 75 0 0 5 0 60 40 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 c 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 T A B L E 2 , APPENTMX. B - 9 0 = 2°m 353222 43553 nr&i 4 45555 25334 :45511 , 9 43333 ogss • 5 0g:s|| 53S| | | * 9 555333 °55532 3223JI " l* §52SS§ oSSSSS SgSSSS 555J| | 9 24553 o2g235 22522 • §S5|1| - 525355 43535 „ 04332 523532 42551 mm » -S5Sgg 555535 s^|H * 033355 553555 33333 • mm » ^ I S a S ^25252 55^11 23SII1 f - 2 £ K * ^ 45353 -SSS|1 5311H s 335525 o2S2S3 2§522| § 5 | | | ! -•oSSSI 4335| 5535|I Kgiii » °5SSriS 4533| 45553 333111 9 04323 553233 43333 S S i i H 4 .§355S -21535 135551 552111 * m*& 45511 ^ i i i i i sssgas sss-sgl f ^ ogggig * °§3S3i 45535 ^ | | | | - 55435 mm mm - "5 Ss S2 . lass g;- S i 5 i g S g§§ 3 - 55453 032233 24333 S S i i i i =« -35355 SasS^l S i l l l l » o555S5 555333 535511 521111 P 323353 02334 5SH11 " ^ -SS35S 45355 SSSHI £833 9 o£55SS 45553 355353 S § | i i i - o25sss 03335s 553323 s a i m - ossaas 45355 533523 s ? s | | | - =23S32 45555 o35522 5232 I I ^ I I 5J |I!i 9 555551 S i l l l ! 035335 °5SSSS §2SS2| § S l | | g " ^ 03453 223333 nnu nun * 23222 • 43331 s l i m - = a S S S 3 o3535S 32S1II s s i i l i - 033232 43332 34332. 2^111 •ntwtw. " - « s > s - a s £ > s - S = a » S " = = S * B pai>j>x»3T « -* - " =. So3S4 222222 S|222S 555522 - zm3 .3323 -ran min => 253522 3.533s ~4n& mm r 4542 S E E mm - .3338 222252 ^li '421% - $m mm 4 4 3 3 mm • 24255 323333 222222 3 2 2 1 1 1 „ 343-2 .2^5 223S2- 4 3551 9 2232-2 022233 -33333 24333 4 23 223:232 332233 33331? » =55232 332322 333333 33343 . a 02343 23432 223333 33333^  " 4 225332 02222J 223322 22223 • * 333223 £52222 43332 2-32|| ' . "33533 45333 433S3.333211 . . 023323 223333 233333 335231 f ^ 4332S 333333 333333 333323 * 253533 223533 43333 222331 » 335335 222223 023333 333311 ^ 43353 333333 43333 333311 " ^ 45353 353533 43331 353111 3 222233 3^5333 o35333 535111 * o33333 335333 2333H 533111 - 233533 o33333 323222 323311 « 5 4 2 3 3 032332 332533 332113 - 43333 o22222 222231 = 355553 233335 35314 . f ~ 023332 323532 533135 2322-^  " l * 232322 023335 323323 33343 s 535352 052255 523551 355111 • 523555 3o5555 54553 535111 - 223233 2253:51 -354 535111 if z. MX P-P. < n UJ % g " n s E = - B -3«s-s - a s s - - : - a g s - s PERCWTACIE lUDUCTiaiO III DJIEAJ\D!G L0AD3 OF TEKJILN JlffiAA TE^T.i Accelerated Weathering Adhaiiva Cycle 2 6 10 11 12 13 14 15 2 0 25.7 3.5 23.9 0 0 . 25.0 0 35.5 23.2 32.t 50.0 31.0 34.6 31.2 57.9 30.5 53.1 1.0.3 t3-9 67.4 35.3 40.7 t5.3 55.1 47.2 35.5 31.7 47.6 62.0 38.1 48.9 64.7 43-0 74.5 67.3 30.0 t9.3 50.e 45.7 56.2 tO.6 53.0 45.1 50.6 36.3 52.5 51.5 43.8 56.0 t5.1 57.0 34.4 60.7 36.9 14.2 0 7.4 t.7 32.6 4.0 10.7 0 tO.3 2i,.0 34.5 28.5 52.6 49.7 43.0 52.6 58.5 t3.7 tO.9 24.1 32.3 t2.2 56.8 76.0 60.4 64.2 45.4 44.2 65.0 29.4 49.7 46,6 44.2 43.5 54.4 t6.6 47.1 35.6 9.0 57.0 22.6 53-2 40.6 76.5 39-1 49.3 41.4 17.6 55.0 40.9 54.7 33.9 64.3 19.t 0 18.5 0 12.7 l . t 9.8 27.6 34.5 31.7 2t.O 50.6 t t .e 35.6 36.9 46.0 U . 2 62.6 to.3 63.3 48.0 1.6.4 52.8 46.4 43.6 49.9 49.4 U . 5 6t.5 44. e 51.4 53.3 42.5 53.2 60.8 84.3 48 . l 66.2 58.1 41.5 51.4 39.2 53.3 38.0 100.0 38.6 99.2 61.0 85.8 53.1 65.4 53.7 54.9 100.0 27.0 34.1 34.3 21.4 23.1 11.8 0 12.8 32.1 tO .3 40.9 77.6 53.9 64.1 58.0 75.7 43.3 80.5 57.3 83.1 80.2 100.0 20.0 86.6 70.4 38.3 100.0 100.0 100.0 100.0 100.0 95.5 100.0 76.1 100.0 100.0 100.0 100.0 100.0 100.0 100,0 100.0 100.0 100.0 100.0 100.0 99.2 100.0 100.0 100.0 100.0 100.0 100.0 100.0 31.6 63.9 73.3 27.1 52.6 98.5 100.0 100.0 6.6 53.4 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 40.4 45.0 73.4 61.4 72.1 12.5 45.6 48.4 63.5 65.6 84.9 11.7 U . 3 77.9 93.5 100.0 100.0 U . 3 100.0 1CO.0 tco.o 1C0.0 103.0 64.8 65.7 58.7 83.7 51.2 45.1 42.7 58.2 70.0 76.1 34.3 50.6 66.6 82.6 100.0 100.0 51.0 100.0 100.0 100.0 100.0 100.0 22.3 43.4 39.5 51.6 U . 7 33.6 57.3 69.9 29.1 59.2 43.7 58.6 100.0 100.0 100.0 100.0 36.6 35.5 40.1 38.4 55.2 55.9 70.6 76.7 86.0 100.0 100.0 100.0 54.1 93.5 100.0 100.0 100.0 100.0 61.7 61.3 39.0 69.7 69.3 55.1 62.0 61.0 51.4 59.8 55.5 69.0 100.0 100.0 39.4 93.4 100.0 100.0 100.0 • 100.0 67.2 47.4 52.9 62.6 55.9 45.0 66.6 52.0 63.5 26.7 56.7 76.9 43.5 97.6 100.0 100.0 100.0 100.0 12.1 19.3 20.5 20.5 20.5 23.1 4.6 10.5 20.5 20.5 20.5 12.5 20.5 25.0 26.8 40.0 40.0 40.0 41.4 29.4 34.1 41.4 40.8 63.4 78.6 85.4 19.0 56.1 71.3 85.4 100.0 100.0 U . 9 36.5 45.4 45.7 41.0 33.7 26.3 31-7 20.6 35.9 41.4 40.3 53.4 42.1 76.5 100.0 100.0 17.6 28.6 25-3 42.0 IB.3' 46.8 99.4 100.0 100.0 20.8 32.1 39.0 57.4 23.1 16.6 18.0 19.1 36.6 20.3 9.6 22.5 18.9 55.7 10.3 35.1 39.6 59.1 91.6 100.0 100.0 11.4 37.9 24.5 34.8 24.1 42.7 31.9 31.5 54.3 37.0 67.8 40.0 52.7 56.7 32.0 66.6 64.5 92.6 100.0 100.0 31.0 30.6 23.9 3.0 11.0 19.0 27.1 31.0 24.5 10.8 16.8 25.0 36.1 51.1 27.4 30.4 20.2 53.0 70.7 17.5 43.2 23.0 23.4 34.5 38.2 34.0 50.6 82.3 62.3 31.2 33.5 36.5 50.1 46.6 58.4 20.5 35.0 40.7 23.1 43.0 45.4 40.0 45.7 61.4 88.7 100.0 57.0 65.8 90.5 100.0 100.0 100.0 16.5 10.1 30.5 50.0 33.e 45.1 48.6 56.7 86.6 100.0 100.0 100.0 69.5 100.0 100.0 100.0 100.0 100.0 22.4 20.3 45.4 22.0 13.3 30.8 15.4 42.3 42.6 66.6 87.6 74.1 38.8 92.6 100.0 100.0 100.0 100.0 31.7 22.5 25.8 22.2 54.6 50.6 5.9 35.9 32.4 65.0 100.0 100.0 1.6 84.6 100.0 100.0 100.0 100.0 19.5 29.9 34.3 31.5 22.7 11.6 12.7 15.1 58.2 64.1 12.0 78.1 100.0 100.0 100.0 100.0 14.5 45.4 21.2 51.7 20.1 18.6 35.7 39.4 37.5 63.9 72.1 58.0 100.0 35.7 100.0 100.0 100.0 100.0 100.0 13.5 10.1 36.7 15.6 46.9 19.8 53.6 29.1 40.1 40.1 66.4 96.6 100.0 100.0 100.0 27.6 30.0 40.6 33.4 51.7 38.4 37.2 42.7 52.0 40.6 35.3 42.1 47.4 54.5 61.6 73.7 100.0 100.0 1CO.0 100.0 H PERCLKTACE REDUCTION., Hi FEU CENT ,*A.D F A I L U ^ CF THijIDH SHEAR TE JT Design l.'u. Ac col or* ted thering Adh»9iT« Cycle T A B L E 3, A P P E N D I X B f; s I - 92 -TABLE 4A, APPENDIX B Comparisons of the Differences i n Predicted Percentage Reductions (in breaking load or % wood failure) Design 11 - Design 11% WF Weath-Adhesive ering Tree No. 15 Cyole 2 8 10 11 12 13 14 I 35.5 6.0 -40.0 46.7 22.3 36.6 -12.7 51.7 II 53.1 21.6 40.4 0 43.4 25.5 61.7 1.6 A III -54.5 63.9 30.0 64.8 39.5 40.1 61.3 ' 67.2 IV -32.7 68.3 73.4 65.7 52.4 9.8 39.0 47.4 V -39.4 74.8 1.4 58.7 53.8 58.8 69.7 52.9 71 -29.3 80.0 2.1 83.7 73.5 70.2 39.3 62.6 I -15.0 -50.0 12.5 51.2 0 0 -5.0 -90.0 II -41.5 27.1 45.8 -34.9 51.8 -41.6 -55.0 55.9 B III -35.8 -47.4 48.4 -7.3 44.7 -44.8 -39.2 -45.0 IV -45.6 -1.5 63.5 -26.8 33.6 55.9 -44.9 66.6 V -23.5 5.0 -9.4 70.0 57.3 -29.4 -13.0 -48.0 VI -15.7 0 -15.1 76.1 69.9 -8.3 81.0 63.5 I 24.5 6.8 11.7 34.3 29.1 -60.3 51.4 -48.3 II -37.4 -46.6 -5.7 -49.4 59.2 -55.6 -40.2 -41.3 C III -50.6 0 -22.1 -33.4 -56.3 -14.0 -44.5 -23.1 IV -15.8 0 -6.5 -17.4 -31.4 0 -31.0 —8.2 V 0 0 0 0 72.8 0 0 -5.5 VI 0 0 0 0 -8.7 e 0 0 I -67.9 -6.3 -55.7 -49.0 -23.1 -45.9 -60.6 -56.5 II -19.5 0 0 0 -10.7 -6.5 -6.6 -2.4 D III 0 0 0 0 0 0 0 0 IV 0 0 0 0 0 0 0 0 V 0 0 0 0 0 0 0 0 VI 0 0 0 0 0 0 0 0 Mean of the differences = *5*5% Standard Deviation of the differences s ±37.36# - 93 TABLE 4B, APPENDIX B « Comparisons of the Differences i n Predioted Peroentage Reductions i n Breaking Load Design 11 - Design 4 Adhesive Weath-ering Cycle 2 8 10 Tree 11 No. 12 13 14 15 I 35.5 6.0 -63.3 34.8 9.6 36.6 7.3 51.7 II 12.2 -3.8 -2.9 -25.4 -2.0 -12.1 -2.4 17.7 III 6.0 7.7 -15.0 20.7 10.4 -9.9 22.8 19.4 IV 24.1 8.7 8.4 9.8 -2.1 2.2 -14.8 -4.8 V 18.8 -6.9 -10.3 9.5 14.7 6.4 5.6 13.8 VI 8.4 0 -1.2 24.4 24.4 10.7 -2.5 23.5 I -2.3 -27.1 II 24.4 -8.3 III 18.8 -12.0 IV 11.2 6.9 V 26.5 0.8 VI -15.7 0 I 32.2 -12.0 II 17.2 11.7 III -9.7 16.7 IV 29.8 2.1 V 50.0 0 VI 6.8 0 I 11.7 0.4 II -8.1 0.4 III 5.6 0 IV 0 0 V 0 0 VI 0 0 12.5 51.2 -5.4 7.5 4.4 2.7 -8.3 -6.5 -8.0 10.2 7.4 -19.1 12.3 12.4 0.9 21.6 18.5 9.9 10.0 30.9 29.1 -4.0 -34.1 21.0 14.6 3.9 -10.8 15.2 13.1 -3.2 6.0 0 -16.3 0 0 -5.1 14.3 29.0 6.9 0.7 6.1 - 7 i l 0 0 8.0 0 0 0 0 0 0 0 0 0 -2.4 -2.6 -34.8 7.4 11.9 23.3 7.6 9.6 1.5 -1.2 8.9 18.8 11.1 13.3 8.5 -9.0 14.3 9.2 0.7 30.9 7.1 -3.2 -9.4 26.2 33.6 1.7 29.1 21.4 10.0 28.8 0 23.1 9.7 0 10.3 17.4 37.4 13.8 15.2 9.2 49.8 2.0 9.5 6.2 2.6 0 0 0 0 0 0 0 0 0 Mean of the differences s *6.5?S Standard Deviation of the differences = +15,42% - 9U -TABLE 5A, APPENDIX B Stannary of Slope Eatios Weath-ering Cycles 1 2 3 4 Designs 5 6 7 11 13 Totals I 9.16 11.83 13.29 3.05 4.23 4.17 6.38 3.85 3.37 59.33 II 11.41 9.08 9.88 7.61 6.60 4.27 7.85 10.68 10.05 77.43 III 14.22 11,82 12.68 12.02 9.18 9.33 10.54 9.75 11.24 100.78 IV 11.08 10.47 11.25 10.16 12.44 11.01 12.21 8.72 9.42 96.76 V 11.77 9.42 12.12 7.85 10.86 9.80 9.10 8.97 11.39 91.28 VI 8.23 10.95 11.40 7.38 6.54 8.49 7.31 7.83 9.57 77.70 Totals 65.87 63.57 70.62 48.07 49.85 47.07 53.39 49.80 55.04 503.28 Means 10.978 10.595 11.770 8.012 8.308 7.845 8.898 8.300 9.173 TABLE 5B, APPENDIX B Analysis of Variance of Slope Ratios Souroe of Variation DP S.S. M.S. Calculated Tabled V.R. V.R. P=.01 Totals 53 368.8970 Designs 8 101.0362 12.6295 3.772** 2.99 Weathering Treatments 5 133.9482 26.7896 8.002** 2.99 Error 40 133.9126 3.3478 Indicates significance at Pa.01 level TABLE 6, APPENDIX B Single Degree Comparis (Single Degree) Comparisons of Designs Totals d 1*2*3 vs. 4*5*6*7*11*13 400.12 303.22 96.90 1 vs. i i 65.87 49.80 16.07 3 vs. 11 70.62 49.80 20.82 1 vs. 13 65.87 55.04 10.83 3 vs. 13 70.62 55.04 15.58 1*2*3 vs. 13 200.06 165.12 34.94 1*2*3 vs. 4*5*6 200.06 144.99 55.07 4*7 vs. 11*13 101.46 104.84 3.38 7.31 s Variance ratio for *4.08 s Variance ratio for of Slope Ratios d2 Div. w c« k MSad2/ok V.R. 9389.6100 18 6 86.9408 25.969** 258.2449 2 6 21.5204 6.428* 433.4724 2 6 36.1227 10.789** 117.2889 2 6 9.7741 2.920 242.7364 2 6 20.2280 6.042* 1220.8036 12 6 16.9556 5.065* 3032.7049 6 6 84.2418 25.163** 11.4244 4 6 0.4760 0.142 significance at P - .01 level, significance at P r .05 level. NOTE.- ;BOARDS 2 THICK, BY A WIDE ARE. TO BE •RIPPED F R O M T H E GREEN A"*A AS SHOWN IN THE DIAGRAM; KI l_N - fcR I ED TO 5l MOISTURE CONTENT; AND PLANED TO MAKE "VENEERS* 0 . 2 0 0 " t 0 . 0 0 5 " THICK. CUTTING PLAN FOR BOARDS FLAT-GRAIN, —EBG El-GRAIN, K. EGX10°(! E D G E . - G R A I N W I T H T H E . C E L L S ' I W A K I N C A N (VNtiLE. O F IO" W I T H . T H E . S U R F A C E . . ) D E C E M B E R 1 9 5 2 1 N O F IG . 1 A P P E M D I X . B 9 - 97 -G L U E L I N E - C L E A V A G E D E S I G N S (NO's1TQ9) Ha4oo"h— S P R I N G W O O D A N D S U M M I £ R W O O C G L U E D A T 9 0 ° A N G L E . ( C R O S S -EiANDE©). DESIGN No.4-. T H E S A M E . A S No.i. , B U T U S I N G E B S £ - G H A , I « V E n e e R S ( W I T H THE. C C L L S A L I G N E D " P A R A L L E L To T H E . S U R F A C E S } . DESIGN No.7 . T n e S A M E A S N o . A . , B O T U S I H G F L A T - G R A I N V E M E E K S , ^ V E N E E R S E D G E - G R A I N ' 5 U T W I T H Tne. C E L L S A T r l O ° T O T H E S U R F A C E . " S P R I N G W O O D - S U M M e R . w o o D B A N D S o F" O P P O S I N G V E N E E R S L A I D U P ^ R A L L E L . ( L A M I M f t T C I i ) DESIGN No.5. T H E . S A m c A s N o . 2 . , E > U T U S I N G E D G E . - G R A I N V E N E E R S ( i e . W I T H C E L L S A L I < 3 < ^ £ D P ^ K . A L L C L T O T H E S U R F A C E S ) . DESIGN No.S. T H E . S A i v \ e A s N o . 5 . , B U T U S I N ^ F L A T - G R A I N V E M E E R S . TENSION-SHEAR DES\GNS tNo'5 I0TO)3^ DESIGN No. 10. V E N E E R S A M D uv A R L I D E N T I C A L W I T H N o . 5 . T w o S E T S A R E K E ^ U I R E D T o " P F I O V H S E - G T E S T S P E C I M E H S , DESIGN No. 12. V G . N B . E H S A M O L A V U P Ane I D E N T I C A L W I T H N O . Q . T W O S E . T S A R E . " R E Q U I R E D To P R O V I U L Q> T E S T S P E C I M E N S . D E S I G N S OF T E S T S P E C I M E N S  F O R C O M P A R I S O N OF T H E I R  S E N S I T I V I T Y * , T H E V A R I A B I L I T Y  O F T H E I R T E S T R E S U L T S . N O T E / . - F O R £ > P E c i r < \ E . N N U M B E R I N G A W D ~ CuTTmc P M T E R H S S E E . F I G . A-. EfietHS E D G E - G K A I M B U T W I T H T H E C E L L S A T I O ° T O T H E S U R F A C E . S P ^ I N C W O O D -O P P O S I H G V £ N t £ R S C T * . O S S E . T J A T A 1 0 " A n G L E - . DESIGN No.£. T H E S A M E A S N O . 3 . , B U T U S I N G E D C E - A IN V E N E E R S ( ije. W I T M T H E W O O D C E L L S A L I G N E D P A R A L L E L To T H E S U R F A C E S ) . DESIGN No.9. T H E S A ( * \ E A s N o . , G . , B U T U S I N G F L A T - G R A I N . V E . N E E R S . ' D E S IGN No.\\. E D L - E - G ^ A I N . V E N E E R S , C R O S S - E I A N I X I D . T W O S E T S A R E . R C C ^ U I R £ D T O " P R O V I D E . G T E S T S P E C I M E N S . DESIGN No 13. F L A T - G R M N \ / F . M F . E R S , C R O S S - E » A N D E O , O T H E R W I S E T n u S A M E A s N o . II. FIG, g. APPEMBIX B - 9 8 -M A R K I N G S Y S T E M F O R G L U L - B L A N K S r>ELCErv,E,fLR- 195*2 PIG 3 A P P E M D I X B lOOl T Y P I C A L L O G A R I T H M I C T R A H S F O K M A H O H S F O R L I H E A R C O R R E I A T I O H A N A L Y S E S D E S I G N i T E S T E D A F T E R I o o D E S I G M 7 T E S T E D A F T E R H I OZ 8 OZ U (Q < m © m o - a M fft D E S I & N li . T E S T E D A F T E K 2 1 . S o u p s I ? . in o m o r-KX) r U J U U J P - 8 0 2 Q 6 0 < U J ( ^ 4 0 u D UJ NURCH1954 F l G. 5 , APPENDIX E> R E D U C T I O N IN B K E A K I N G L O A D , ( P E R C E N T ) 101 -HUGHES OWENS 11 * B F I G . 6, A P P E N D I X B 100 AO H U G H I S O W E N S ] K i FIG. 7, APPENDIX B - 104 -

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0106855/manifest

Comment

Related Items