of Terrestrial Algae. II. 
689 
B. Description of Metpiods adopted in the Subsequent Investi¬ 
gations and Discussion of Possible Sources of Error. 
In view of the facts which have just been detailed it will be evident 
that little can be gleaned from comparative estimations of the minimum 
concentrations necessary to effect plasmolysis, since such concentrations are 
only to be related to a small percentage of the cells. A better comparative 
method of obtaining some conception of the relation to plasmo lysing solu¬ 
tions seemed to be to employ a solution rather above the minimum 
and to determine by counting under the microscope the percentages of 
plasmolysed and unaffected cells. In order to obviate changes of con¬ 
centration during the estimations a ring of marine glue dissolved in xylol 
was run round the edge of the cover-glass. Usually at least 1,000 cells 
were counted at each estimation, but occasionally this was impossible 
owing to lack of time ; in some cases the result seemed so even that a 
smaller number of cells was deemed sufficient. As a general rule from 
fifteen to forty-five minutes were occupied in an estimation. The estima¬ 
tions were made with the help of a mechanical stage, and usually the 
material under the cover-glass was counted exactly as it appeared with 
the movement of the stage, thus obviating any personal bias. 
The sources of possible error in this method of estimation are: 
(1) lack of uniformity of material, (2) personal error in the counting, and 
(3) effect of the xylol in the sealing medium on the permeability of the 
material. The error under items (1) and (2) was determined by repeated 
estimations of different mounts of supposedly identical material, and the 
full results are epitomized in Table IV. In general the amount of error 
will be found to be small in relation to the percentages upon which any 
stress has been laid in the subsequent matter. It will be evident that more 
reliance can be placed on the bigger totals than on the smaller. The 
only serious discrepancy among them is that shown by the first count 
for Hormidium , but we doubt very much whether an error of this magni¬ 
tude has often entered into our estimations. In making the latter it was 
noticed again and again how the totals for the different categories main¬ 
tained the same relative proportions during successive periods of the count 
(cf. also the analogous results often obtained with the same alga from the 
same habitat at different times ; see Table VII). A very striking uniformity 
is shown by the data for Pleurococcus, which is no doubt due to the possi¬ 
bility of getting a better assortment of such a unicellular form than can be 
done in the case of the filamentous types. 
The possible error introduced by the presence of xylol in the sealing 
medium was investigated by comparing the counts for material sealed with 
melted marine glue and with marine glue dissolved in xylol respectively. 
z z 2 
