[Vor. 8 
352 ANNALS OF THE MISSOURI BOTANICAL GARDEN 
standardization purposes. It was, however, with some regret 
that the use of gold sols was then considered undesirable, since 
the sizes of the particles in such solutions have been so well 
determined. In undertaking the standardization work it seemed 
best to use at the outset colloidal solutions that might represent 
extremes in sizes and then to narrow the field down to those 
that might correspond more nearly with the particles of the 
mosaie disease. Accordingly, a solution of dextrin was first 
used, since the particles represent extreme smallness in colloidal 
solutions, and moreover the filtrates could be readily tested by 
the simple iodine method. Filtration experiments with a 1 per 
cent dextrin solution indicated that these particles passed freely 
through all of the standard unimpregnated porcelain cups 
employed. A small quantity of dextrin passed the cylindrical 
cup C, impregnated with Al(OH), and none passed the other 
cup so impregnated. 
In the next test milk free of fat was employed with a view to 
determining the size relation between the mosaic disease par- 
ticles and casein in milk. The milk was first filtered through the 
spherical atmometer cup and it was found that this filter prevents 
entirely the passage of casein. The filtrate was a clear solution 
containing no demonstrable quantity of casein. It was now 
necessary to utilize a larger colloidal molecule for standardization 
than dextrin| and yet a molecule considerably smaller than 
casein in milk, thus hemoglobin was selected. 
The hemoglobin employed was a preparation made by standard 
methods from ox blood. As soon as the ox blood was drawn 
neutral potassium oxalate to make 0.2 per cent was added in 
order to prevent clotting. A measured quantity of the blood was 
then distributed in centrifuge tubes and centrifuged, the 
supernatant serum being drawn off and the known volume of 
corpuscles thoroughly washed 4 times with a physiological salt 
solution (0.9 per cent NaCl) An equal volume of distilled 
water was added to lake the corpuscles, after which the solution 
was again centrifuged to remove fibrin and stroma. The red 
supernatant colloidal solution was finally diluted so as to contain 
1 per cent hemoglobin, estimating the original hemoglobin 
blood content at 12 per cent. For this work it was not necessary, 
