22 



DISPLACEMENT INTERFEROMETRY BY 



other series were now worked out, of which figure 21 (2) shows an example. 

 The mean rates per turn were from A<?=y.7 to Ae = 6.2, corresponding to 

 module from E X i cT 11 = i . o to i . 2 5 . I also resorted to the method of observing 

 in triplets between definite steps of pressure. The results were all of the same 

 low order. 



With a more robust interferometer the series of results in figure 22 were 

 obtained. The outgoing and return data are here more nearly coincident, 

 but the graphs are not as a rule straight. The mean rate may be taken from 

 figure 22 (i) and is per turn A = 8.6. The fringes were of moderate size, so 

 that 



io- u E= i2.75/Ae = 12.75/8.6= 1.5 



Larger fringes were now installed, giving A0/A./VXio 3 = 34.8. The results 

 are shown in figure 23 and are again nearly coincident, but lie on curved loci, 

 often with two definite rates. In the first series the larger is A? = 8.4 per turn, 

 in the second series A = 10.0 per turn. Hence 



io- u .E=i6.i/Ae=i.9 and 1.6, respectively, 



larger than the preceding, probably because the face had been reground. But 

 this order is still only about one-third of the normal modulus of glass. 



The endeavor was now made to proceed as in the case of brass above, with 

 a shouldered rod and thinner sections. With this in view, the glass rod was 

 fixed in a small hollow cup (fig. 7c) with fusible metal. The cup being threaded, 

 was thereupon screwed into the cross-piece A, figure 7. The first glass rod 

 was L= 1.9 cm. long, 0.32 cm. in diameter, A =0.083 cm - 2 Very small fringes 

 were used, Ae/A^V Xio 3 = 13.3, so that XicT 11 = 6.5/A. The results in 

 figure 24 conform to Ae = 3-4 to 4.2 per turn of screw, which correspond to 

 E X io~ n = i .9 to 1.5. This is therefore no improvement. 



4 



5 fyuM Z 3 4-tu*t& 



The rod was then drawn thinner, viz, to L=i.g cm., diameter 1.9 cm., 

 A =0.284 cm. 2 The results in figure 25 are necessarily confined to small 

 stresses and show a gradual improvement. As Ae/ANXio 3 is here 14.2, EX 

 io~ n = ig.i/Ae. &e per turn lies between 6.2 and 6.8, so that .EXio"" 11 is 3.1 

 to 2.8, respectively. Experiments made in triplets gave Ae~6 or EXio~ ll = 

 3.2. The discrepancy is undoubtedly connected with the difficulty of securing an 



