23 



strength with moisture was developed entirely free from other influences. In many cases a second 

 assumed curve was drawn and a new diagram of results obtained. (Fig. 11.) 



Notwithstanding this method of elimination of other sources of strength it was found in sev- 

 eral instances, as in the modulus of elasticity, crushing across the grain, and in shearing (Plates 

 VIII, IX, and X), that some of the trees showed such discrepancies in their relation of strength 

 to iiioisture, that the curve of mean values could not be used for reducing results on these trees. 

 Such discrepant sets of results are inclosed by dotted lines, and separate curves of correction 



3000 PERCENTAGE OF MOISTURE^ 



Fig. 11.— Method of coiistmctiiig ciuvos of .averages. Long-le.af pine (Phuis polustris). 



employed in tliese cases. The corresponding pairs (dry and green) of curves for these trees can 



be ide^itilied by the attached numbers of the trees, the average values for which arc re])rcsente(l 



by the jilotted points. 



Oh those sticks which were originally 18-feet long beams three sets of results were obtaincil 

 by tests at approximately .iO, 20 and 12 per cent moisture. The very high percentages of niois 

 ture shown in all these <-urvcs do not represent averages for the tree, but represent the mean re- 

 sults from such sticks as iiad a very liigh pro])orti(m of moisture in tlic grc-u tests. They are 

 inserted to give some further indication of the form of this end of tlie curve. 



