THE AID OF THE ACHROMATIC FRINGES. 61 



The cycles remain, though they are more slender than before. The general 

 trend of the observations, indicating a continual slow viscous yielding through- 

 out, superposed on the hysteresis, is the same as before. The change of the 

 modulus with the load, i. e., 



P/A= 18 29 41 53 kg./cm. 2 

 io~ 10 =i.64 2.30 2.83 3.00 



is in the first three cases not very different from the preceding. The fourth 

 datum is low. One may write R = E/A(P/A) = io 10 Xo.o5. It is question- 

 able, therefore, whether lateral support received from the rigid sheath can 

 account for the increment of E with the load. The variability of R is rather 

 due to the occurrence of hysteresis, whereby the datum for E is very variable 

 unless found in triplet observations between two definite loads. A few 

 additional triplets (3 or 4 for each step of loads, added to elucidate this 

 question) follow: 



P= i to 2 2 to 3 3 to 4 4 to 5 kg. 

 P/A= 18 29 41 53 kg./cm. 2 



Mean io l E= i.SS 2.84 3.44 3.32 



These data are all larger than the preceding set, owing to the different 

 method of observation (triplets confined to two fixed pressures and not to 

 pressures varying cyclically over a large interval); but the general trend 

 of results (see fig. 436) is the same as before. As E is a maximum in 

 the interval between 3 and 4 kg., it does net seem probable that the possible 

 sustaining effect of the walls of the sheath can have had any important 

 influence. More probably complications of vicosity, hysteresis, and possibly 

 temperature, or even of adjustment, account for the erratic behavior ob- 

 served. As such I have refrained from pursuing it further. 



32. The same. Brass. To estimate how far the above work may be in 

 error, owing to inadequate rigidity of apparatus, the hard-rubber rods were 

 replaced by brass rods of about the same size. Their constants were 



L = 2.35 cm. 2r = o.375 cm. A=O.IIO cm. 2 R=io.^ cm. 



R' = f.o cm. * = 4S 



An example of cycles obtained in this way (after initial loading) may here 

 be given: 



Load 123 4 5 4 3 2 i kg. 

 io 4 AAf o 6.5 ii. o 12.0 12.5 10.5 9.0 7.5 1.5 cm. 



One may infer that below 2 kg. there is some dislocation, but very little 

 is shown by the residual loop above 2 kg. Moreover, the loop is not a case of 

 hysteresis, as is clear from the inverted march of values. Hence the mean 

 value AA/YAP = 0.00016 cm. may here be accepted, from which E= 5.5 Xio 11 

 may be computed. This result is too low; but with a micrometer reading 

 to but io~ 4 cm., limiting the displacement interferometer, a value even as 

 good as this is hardly expected. It indicates the degree of deficient rigidity 



