THE COLLAR-CELLS OF HETEROCGLA. 35 
The difference of the best two series of sections from all 
the others is in the uniformity of their contraction. It will 
be seen from the drawings that while the living form of the 
cell is barrel-shaped (figs. 1, 2, 3, 19), the tendency of preser- 
vation is to produce a sphere (figs. 17, 18, 20, 21; Dendy’s 
figs. 24 and 25, plate 14, vol. xxxv, and fig. 38, plate 4, vol. 
xxxii, of this journal, &c.). This necessarily produces a highly 
disproportionate contraction at the base of the collar and it 
results that the measurement of the ratio of this dimension to 
the greatest width of the cell affords a fair index of the distortion 
which the preparation has suffered. Thus the artifact nature 
of Sollas’s membrane is concisely demonstrated by the follow- 
ing figures, averaged from all the measurements : 
Basal width in living cells ‘ . Orap. 
in balsam with separated sila, e aaoy cs 
s in balsam showing Sollas’s membrane 3 AO: Gps: 
Collar-width in living cells : vs xs Olbs 
35 in balsam with separated eallare. 2, Aeon. 
in balsam showing Sollas’s membrane nye ah pee 
In life, as in the preparations where collars are separated, 
the collar-width—that is, the apical width of the cell—averages 
three fourths of the extreme width. Where Sollas’s membrane 
is present the collar-width ranges from two thirds to one third 
of the extreme width of the cell. 
The change can be best followed by comparing figs. 19 and 
20 (series B), which are drawn from the same sponge to the 
same scale,—the one in life, the other from a paraffin section 
mounted in Canada balsam. 
The nett result of the measurements may be seen in the 
averages of three series of paraffin sections, D, C, and A: 
cubic p. B 
Cell-volume in life (ef. figs. 1, 19) . 270 ~ Collar-width in do. . 4°6 
is in balsam, Series D (ef. fig. 15) . 170 5 B 4°3 
% i elie. . 185 i i 3-0 
B 3 3 Aneities.. 17, 
18) . 865 K ‘ 2-9 
There are, therefore, two principal phenomena due to the 
