36 GEORGE BIDDER. 
transference of cells through osmic acid, alcohol, and benzol, 
into paraffin, and finally Canada balsam : 
(1) There is a reduction in the total volume of the cell, 
which apparently cannot be avoided, corresponding to a mean 
linear contraction of about 5: 4 in the best preparations, and 
5 : 3 in the worst. 
(2) Independently of the extent to which this takes place 
there is generally a change of form. It appears possible (cf. 
figs. 1, 15) almost entirely to avoid this, but by most methods 
the rectilinear and angular outlines of life (figs. 1, 2, 3, 19) 
are replaced by pyriform (figs. 20, 21), ovoid (fig. 17), spherical 
or even oblate (fig. 18) contours in the permanent preparations. 
Thus, taking from the averages of the last table the con- 
sequent ratios of the linear dimensions to those of life, we 
obtain : 
Mean linear contraction ratio Contraction ratio of collar- 
in— width in same sections. 
Series D (fig. 15) . 82 : 2 360 
Series C : Sas) : . 68 
Series A (figs. 17,18) . ‘6 ; . 46 
It was experimentally shown that the extreme changes of 
cell-form were not produced in alcohol. Bringing part (E) of 
a sponge in four minutes through 30 per cent. and 50 per 
cent. into 70 per cent. alcohol, the cells were compared in 
paraffin sections with the part (D) of the same sponge 
treated uniformly by slow dialysis. The collar-width (4:0 ,) 
and the basal width (5°44 to 5°7u) in E retain their normal 
proportions to each other, and the collars are not united. It 
is true, however, that the mean contraction is greater (ratio 
‘74) than in D, and the height of the cells is disproportionately 
diminished (7°8 « as against 9°5 w in D and 12:0 w in life). 
It was also experimentally shown (fig. 16) that in some 
sections of the best series (D, cf. fig. 15) stained on the slide 
in the ordinary way through turpentine and four grades of 
alcohol into Grenacher’s hematoxylin, the cells suffered con- 
siderable distortion, and in many cases developed Sollas’s 
