PHYSICS: C. BARUS 
47 
taken to repeatedly tap the movable parts of the apparatus before each 
reading, and these came out remarkably smoothly at once. 
An example of experiments made after some improvements of apparatus 
and carried out with the rod specified through a range of about 15 kg. is given 
in figure 5, a. The rod stood the stress well except at the end where it showed 
slow viscous shrinkage. The data (ordinates) contain the running displace- 
ment A TV of the micrometer screw in terms of the successive turns of the 
forcing screw, iv, figure 1. Stress increasing or decreasing is indicated by 
arrows. 
The values of E were computed from 3 turns (5.46 kg.) of the forcing screw. 
For loads up to 5 additional turns (total 11 kg.), the data for E are practically 
identical, both in the outgoing and return series. See figure 5, h. At 6 
turns (total 13 kg.) the rod apparently yields; but at 7 turns it again stiffens 
in both cases. As a whole the data are quite as good as the reading of the 
micrometer screw admits. To interpret this apparent increase of E it would 
be necessary to use a thinner rod, as the following experiments with brass 
and glass suggest. Again, only in case of more rigid rods, where A N fails, 
is it necessary to use the ocular micrometer (A e). 
4. The same. Brass. — By way of contrast, a thick solid brass rod, L = 
2.34 cm. long, 0.376 cm. in diameter, area A = .111 cm.^ was now put into 
the sheath, s, figure 1, and tested, the aim being to redetermine the limit of 
measurement. Here n / A e or the ocular micrometer is the essential datum 
and A e / A N must be known. 
The interferometer was modified to guard against displacements due to 
tremor, large fringes were installed and readings were made several times 
before and after tapping. There was but little difference. An example of 
such results is given in figure 6, (1) and (2), where A g / A iV X 10^ = 43, 
and therefore per turn of screw (n = 1), E — X 19.3 A e. 
In the first series A e per turn was 8.0 and hence E X lO^^^ = 2.4; in the 
second A e = 7.8 and E X 10"^^ = 2.5. Seeing that a scale part of A ^ in 
figure 6 is but 23 X 10~^ cm., these results are experimentally very good; 
but their absolute values, as given by E, is nevertheless very low. The 
rates for the outgoing and return series are identical. The backlash, as it 
were, on passing from one to the other is probably in the apparatus. 
In triplet observations, naturally, higher values of E will be found; for 
instance, between 3 and 4 turns of the screw, A e = 5.5 per turn, appeared 
in successive independent experiments. Thus E X 10"^^ = 3.5. 
An example may now be adduced of experiments made with a brass 
rod, thin and shouldered as indicated in figure 3, the large end (J inch in 
diameter) being threaded and screwed into the cross-piece A. The dimen- 
sions of the thin part were, length L = 1.8 cm., diameter = .22 cm., A = 
.038 cm.2 The fringe factor was A g / A X 10^ = 29. The mean rate per 
turn was found to be A e = 4.9 and 10 X = 29.2 / 4.9 = 6.0. 
