over into the second test tube with them, thus nullifying one of the most important features of the method, i. e., the getting rid of the sand. In the work of the Board, clean white quartz sand known as Berkshire is used.

In order to test the filtering efficiency of different sizes of this Berkshire sand a series of experiments was undertaken, the results of which are given in the following table. By "size of sand 40-60" is meant that size of sand which passes through a brass sieve having 40 meshes to the inch, but which does not pass through one having 60 meshes to the inch. The experiments were duplicated for each size of sand.

To make the experiment conclusive distilled water was used, to which had been added common yeast containing some starch grains. Yeast was used because its cells are smaller than any of the microscopical organisms found in Massachusetts drinking waters. If the sand retains yeast cells it is safe to conclude that it will retain larger forms as well. The number of cells which had passed through the sand filter was found by refiltering the effluent through the very finest sand (finer than 140) and examining by direct inspection of the sand, according to the original Sedgwick method. In this method a very small amount of sand is used, and, after filtration, sand and organisms are washed into the counting chamber where 20 fields are examined. (Report on the Biological Work of the Lawrence Experiment Station, loc. cit. page 806.)

The decantation error or error due to organisms remaining in the sand after decanting is determined as before, by washing the sand with a second five cubic centimeters of distilled water. The efficiency of the different sands is found by adding together the percentages of yeast or starch observed by the usual method, and the percentage of the decantation error.

TABLE IV. Showing the Filtering Efficiency of Different Sizes of Sand for Yeast Cells and Starch Grains.

NUMBER OF
CELLS OBSERVED

BY

USUAL METHOD.

SIZE OF

SAND.

TABLE IV Showing the Filtering Efficiency of Different Sizes of Sand for Yeast Cells and Starch Grains - Concluded.

Number.

NUMBER OF
GRAINS OBSERVED

BY

SIZE OF
SAND.

USUAL METHOD.

B. STARCH GRAINS OBSERVED.

NUMBER OF
GRAINS OBSERVED

AFTER SECOND
DECANTATION, OR
DECANTATION

ERROR.

NUMBER OF
GRAINS OBSERVED
AFTER SECOND
FILTRATION, AND
THROUGH

NUMBER 140 SAND.

1

1

40-60

1,250

95.57

From this series it is seen that the efficiency increases as the size of sand diminishes, and that sand which passes through a sieve of 60 meshes to the inch but not through one with 120 meshes to the inch is the most efficient of those used. The efficiency for starch is greater (98.8 per cent. for the finest sand) and the difference between the sands for starch is not so marked as in the case of the yeast, because of the comparatively large size of starch grains. It will be noticed that the percentage of yeast observed by the usual method is greater for the sand 80-100 than for the sand 60–120, because of the large decantation error in the latter case (25 per cent. for yeast and 18 per cent. for starch). This decantation error remains otherwise about the same for all sizes of sand, and I am at a loss to account for the large percentage in this particular case. I have never been able to find as great an error from this cause with this sand at any other time; and Table III. shows that the average decantation error for total organisms is only 7.53 per cent. for the same size of sand which has been used since March 1, 1891. This 60-120 sand is composed of sand of different sizes, in the following proportions: 60-80, 39 per cent. by weight; 80-100, 34 per cent; and 100-120, 27 per cent.

A COMPARISON OF TWO DIFFERENT PORTIONS OF THE SAME WATER EXAMINED SUCCESSIVELY BY THE SEDGWICK-RAFTer Method.

A valuable and practical test of the method consists in the comparison of the results of examinations of two different cubic centimeters out of the five, and also of two different portions of the same water under precisely the same conditions.

The following experiment was undertaken for the purpose of comparing the results of separate examinations of two cubic centimeters of the five. The two portions were withdrawn from the test tube one after the other and examined. The results are shown in the following table, in which the figures represent the number of organisms in one cubic centimeter of the original sample.

Showing the Variation in Numbers of Organisms Observed in Two
Different Cubic Centimeters out of the Five.

These portions were examined independently of each other and the results were not known until this entire investigation was completed. It will be noticed that the totals are higher for the second cubic centimeter examined, in every case. This increase in totals may probably be explained as follows: the one cubic centimeter taken for examination is usually withdrawn from about midway between top and bottom of the five cubic centimeters of decanted liquid after this has been thoroughly shaken. In the time elapsing between shaking and the withdrawal of the one cubic centimeter for examination, the organisms settle more or less, giving a certain concentration of organisms in the four cubic centimeters remaining in the test tube.

The following experiment was undertaken for the purpose of comparing the results of examinations of two separate portions of the same water. Two equal portions were taken from the same bottle and allowed to filter through equal depths of similar sand. The experiments were made by the same observer upon five different waters and the results are given in table VI.

TABLE VI. Results of Examination of Two Equal Portions of the Same

* Rhizopoda.