is, 5 Valencia: Commercial Philippine Timbers 503 
115 per cent of the average; in the case of apitong (fig. 12), 
22 per cent of the tests fell between 110 and 115 per cent of the 
average. 
Had the tests been more numerous there would not be gaps 
between the cross-hatched areas in figs. 11, 12, and 14, and the 
diagrams would be more symmetrical with respect to the median 
line or the line showing average modulus of rupture. The dia- 
grams would be very short at the ends and highest at the middle, 
somewhat like fig. 13. Small specimens are less variable in 
strength than the structural sizes. The reason for this is that 
the small specimens are free from defects. 
Reference number. 
Figs. 15, 16, and 17 show the relation between dry weight and 
the modulus of rupture obtained from tests on small clear beams 
free from defects. Figs. 18 and 19 show the relation between 
dry weight and modulus of rupture, based on results obtained 
from tests on large beams. In general it will be observed that 
the dry weight has a distinct tendency to increase as the modulus 
of rupture increases. Where a heavy timber of structural size 
had a low modulus of rupture, it will generally be found, by 
referring to the tables of results in the appendix, that the timber 
had serious defects on or near the tension face which in all 
probability greatly weakened it. The strength of specimens hav- 
ing the same dry weight sometimes varies considerably. This 
