Voiv. 6, 1920 
PHYSICS: A. A. MICHELSON 
125 
The introduction of instead of t itself is a step so radical that it may 
be well to give an illustration in its justification. 
For this purpose it is desirable to choose a material in which the elastico- 
viscous effect is well marked. This is notably the case for vulcanite, which 
has the added advantage of the relatively small importance of the third 
or purely viscous term. This illustration is, perhaps, the most striking 
in showing the appropriateness of t^^^ instead of t; but all the materials 
investigated give similar results. 
Following is a table of results for Ro , the return at the time t after re- 
leasing the stress.^ ^V^ gives the result of calculation from 
R = 890(1 — ^-•57Vr) 
Rt gives values calculated from 
R = 840(1 — ^--^O. 
The differences between calculated and observed values under Ai and A2 
show that the former expression is very near the truth, while the latter is 
entirely inadequate. 
TABLE III 
t 
R 
0 
Rt 
Ai 
Ri 
A2 
1 
380 
387 
7 
277 
—103 
2 
490 
492 
2 
462 
—28 
4 
600 
605 
5 
672 
4-72 
9 
730 
729 
—1 
820 
+90 
16 
800 
802 
2 
838 
+38 
25 
840 
841 
1 
840 
00 
30 
853 
851 
—2 
840 
—13 
00 
890 
890 
0 
840 
—50 
While the term involving a permanent set may not have any applica- 
tion to the problem of the Earth tides, yet it may not be amiss to draw 
attention to the fact that in some cases and especially at temperatures 
approaching the melting-point this term becomes the most important of all. 
The temperature coefficient in this case enters in the form Q/T — 9 ; 
giving as it should perfect fluidity at T, the melting-point. ^ 
In the former article the expression given for this viscous term is 
53 = {Fty in which F = CzPe^^^ and p is stated to be approximately 
one-half. 
From more recent data the average value of p is 0.41; and if from the 
nineteen substances examined, four be excluded the average is 0.35 which 
makes it much nearer one-third than one-half. 
The expression for the viscous terms should be = (Ft)'^^ if the stress 
(P) is constant. If P is a function of time 
53 = (fFdt)'^: 
