﻿190 C. Barus and V. Strouhal — Strain-effect of sudden 



If we construct the density of consecutive shells, d, as a 

 function of their mean radius, R, and then compare the dia- 

 grams which obtain for the five P. K. drops (Nos. 3, 5, 7, 8, 9) 

 we find that the loci have no salient feature in common. Hence 

 the true variation of d from surface to axis of a P. K. drop is 

 here unrecognizably obscured by errors of observation. In- 

 deed we shall find below that the probable total variation of o 

 attributable to strain will not far exceed 0*5 per cent. The 

 mean accuracy of d is certainly not much within this figure. 

 Again if we compare the contours of corresponding curves, 

 d=f\R\ in Table V for drops free from strain, with each other 

 and with the contouis belonging to Tables III and IV, we 

 encounter fluctuations of the same kind in both cases. These 

 errors include the inaccuracies of mere measurement (the mass 

 of the drops is unfortunately small) as well as such discrepan- 

 cies as result from the unavoidable invasion of small bubbles 

 during solution. 



If, however, we compare the mean values of d for strained 

 shells (Tables III, IV and VI) with the mean values of o for 

 shells free from strain (Tables V, VI), we find the latter values 

 (unstrained glass) always in excess of the former. In other 

 words, although our results are insufficiently sharp to enable 

 us to describe the exact nature of the temper-strain in glass, 

 they do permit us to classif}^ it as a strain of dilatation, so far as 

 we have observed, throughout the substance of the drop. . This is 

 an inference of importance and we have therefore drawn up 

 the following general tabular comparison, to supplement the 

 special and direct comparison given in Table VI. 



In Table VII, J s is the density of the glass itself after thor- 

 ough annealing at red heat, as found in our earlier paper," — 

 the datum being the mean value for six drops. J/ is the dens- 

 ity of the glass after fusion in a platinum basket and very slow 

 cooling. The additional increment of the density of the glass 

 thus produced is to be noted. o h and o' denote the densities of 

 shells, strained (quenched) and unstrained (annealed at 450°), 

 respectively. The sixth column contains the relative value of 

 decrement of density for each drop; the second column the 

 number of shells whose average d is the datum given. 



To facilitate further comparison we insert also the corre- 

 sponding table for steel. f Here 2p are the diameters of the 

 steel rods, A hy J s , J c , their densities in the hard, the soft and 

 the commercial (soft) states, respectively. J\ finally, denote the 

 densities of steel at the end of the first phase of annealing 

 (Ca 350°) and therefore apply for rods free from strain. The 

 table gives us the following relative decrements of density : 

 column first, the total decrement ; columns second and third 



* L. c. p. 449. f Bull. U. S. G. S., Xo. 27. 



