176 CAREY: GROSS STRUCTURE OF AN AGAR GEL 
From the larger piece, which was dried to constant weight, the 
per cent of water in the gel at the given time was obtained. This 
gave a fairly accurate estimate of the relation of water content to 
structure but cannot be regarded as strictly quantitative, as the 
electric heater did not give a wholly uniform temperature over all 
the drier so that when the time factor is constant a slight difference 
in the position of the strips when in the oven might vary the 
concentrations in them. 
The tables on page 175 show the per cent by weight of water 
and agar in the gels: (@) when there was no structure present; 
(6) when it first began to appear; and (c) when the structure was 
complete. From comparison of TABLEs I and II it will be seen 
that the critical point for structure formation lies between about 
g2 per cent and 92.5 per cent water content. This, however, is 
not the uniform concentration throughout the piece, since the 
surface would of course lose water first and the water content in 
the center of the gel would be higher than that on the edge, as 
it dries from the outside in and from the edge of the frame toward 
the center. The structure, as one would suppose, starts from the 
outside of the block and goes inward. This would necessarily 
be the way since the surface loses water first. The structure 
was complete, as shown in TaBLE III, when the gel contained 
about 21.5 per cent by weight of water. 
For the purposes of this paper the agar dried to constant 
weight in the desiccator was assumed to contain no water. The 
basis of the calculations was as follows: the per cent of water by 
weight to the total weight of substance was calculated by sub- 
tracting the final (constant) weight, obtained by drying the 
pieces of agar in a calcium chloride desiccator, from the first 
weight, after drying varying lengths of time in the oven. The 
loss of weight was therefore the per cent of water in the gel when 
structure became apparent. The pieces gained somewhat in 
weight while in the desiccator due to the vapor pressure of the 
calcium chloride. This was not taken into account as it is a 
_ constant error in all cases. 
The structure varies in the different gels but is fundamentally 
the same. It appears as horizontal splitting in the gel, and when 
very pronounced as it was in most of them, particularly those 
