282; Proceedings of Indiana Academy of Science, 
to attach themselves to surrounding crystals. As the crystals build up there 
is a shrinking in the length of the specimen. This shrinking continues until 
the more easly occupied spaces are filled, the displacement gradually be- 
coming less and less until it is not detectable. But there is still strain left 
for not all metals anneal perfectly at ordinary temperatures. When more 
strain is produced by applying stress there is an agitation of the particles 
of the metal and the shrinking starts again, as soon as the stress is re- 
moved. Since in the drawn wire a large per cent of the metal is in the 
amorphus phase it is only logical to expect that there would be a greater 
recovery for a given immediate strain than in an annealed specimen. 
It is easily seen from this viewpoint how increased stress and increased 
time of applying stress produce greater recovery. Starting with an annealed 
specimen, the greater the stress applied the more erystals there are broken 
down and the more amorphous substance there is to take part in the process 
of crystal formation, hence the greater contraction. The same argument 
holds for increased time of applying stress. 
There is no legitimate basis of comparison of the rapidity of contraction 
of two different metals. A suspended aluminium wire a meter long meets 
but comparatively little opposition to contraction due to its own weight. A 
piece of lead wire a meter long suspended by one end, when freshly annealed 
flows of its own weight. This indicates the great force that must be over- 
come, in the case of lead, by the forces of recrystallization, even to main- 
tain the original length. Since experimental results show that there is actu- 
ally greater recovery for lead per unit of length per unit of stress applied, 
other conditions being the same, in spite of this handicap, than for either 
copper or aluminium we see how much greater must be the forees that 
rause the shrinkage in lead. But lead anneals perfectly at ordinary tem- 
peratures, aluminium at higher temperatures and copper at still higher 
temperatures, just the order that must be expected if recovery is to be 
accounted for by recrystallization. The fact that greatest recovery takes 
place where greatest activity of recrystallization is involved is a strong 
point in favor of the hypothesis that the one is dependent on the other. 
This idea fits exactly Prof. Michelson’s (5) picture of elastico-viscous 
recovery. The force that causes the shrinkage is an elastic force but pro- 
duces no instantaneous effect for just the same reason that a rubber band 
stretched on a block of wood cannot contract to its original length. But 
eause the block of wood to contract gradually by any means whatsoever 
and the rubber band follows it. In just the same manner the elastie forces 
which are contained within the remaining crystalar structure cannot act 
because they encompass the amorphous phase of the material. But let this 
phase begin to reform into crystals. It is wedged between the crystals 
and fills all the spaces between them. As it joins neighboring crystals or 
forms new ones the original crystalar structure begins to make a readjust- 
ment because of the strain which it is under. The more active the amor- 
phous phase is the more rapidly the whole structure contracts. 
Such a conception of the state of a metal after strain will account for 
what Prof. Michelson (5) calls “Lost Motion’, the failure of a strained 
metal to return to its original configuration when the stress is removed. It 
is found that the more nearly perfect the process of annealing is, the greater 
