22 



tr. S. BUEEAU OF nSHERIES. 



ing any particular sample one time through the average amplitude, 

 thi'ee-fourths radian, or 42.97°. Of course, since each oscillation 

 is of different amplitude, the amount of energy that is transformed 

 with each oscillation is accordingly smaller than that for the preced- 

 ing, but for practical purposes only the average would be necessary. 



It may be objected that the friction of the air is considerable and 

 variable. Efforts were made to determine tliis source of error, but it 

 was found impracticable to do this precisely, because the angle chosen 

 was too large for the swinging of a frictionless (agate knife-edges) 

 bearing. For this reason a smaller working angle might be better as 

 permittmg an accurate determination of the air-friction error. In 

 any case this error is small and practically constant, so that it is not 

 corrected in this paper. 





12 3 4 



NUMBER OF MONTHS LXPOSED 



Fig. 11.— Flexibility of cotton lines exposed in sea water at Beaufort, N. C. 



Two other sources of error in this test require attention. One is a 

 variation in the flexibility at various places on the sample, the other is 

 a change in flexibility caused by the action of heat, generated by bend- 

 ing, on the viscosity of the preservative. The method of making the 

 test takes these two factors into account. Ten pieces, each about 32 

 cm. long, were cut from the sample to be tested. One was placed in 

 the jaws of the tester and the number of oscillations counted in the 

 manner described. This test was repeated 10 times at the one place 

 on each test piece, 100 tests being made on each sample. If heating is 

 considerable the successive tests at the same place will show increasing 

 flexibihty with preservatives that are subject to variation in flexibility 

 with temperature. Progressive increase in flexibility on repeated 

 bending was found in lines treated with tar, petroleum product No. 1, 

 gilsonite, copper oleate, slight increase in quercitron and potassium 

 bichromate, the Dutch method, petroleum product No. 2, and copper 

 paint No. 2, and no increase at all in copper paint No. 1 and untreated 

 white line. 



