bubble in the glass itself, situated a few millimetres above the mark, was taken as a more convenient standard of reference; the distance between the lowest part of the bubble and the best defined limit of the adjustment mark was measured once for all. The water level was then adjusted so that the top of the meniscus was at this ascertained distance, and a weighing of the water contents made (temperature 24·2); in subsequent experiments the water level was purposely adjusted 1·6 mm. below and 1 mm. and 2 mm. above the standard mark, both series being required as the capillary limb was not a true cylinder.
The following results were obtained:—
| No. | Level. | Water contents at 24·2 (corr.). | Difference. |
| I | 0 | 48·9402 grammes | — |
| II | 1·6 mm. (below) | 48·9373 grammes„ | −0·0029 gramme |
| III | 1·0 mm.„ (above) | 48·9410 grammes„ | +0·0008 gramme„ |
| IV | 2·0 mm.„ (above)„ | 48·9428 grammes„ | +0·0016 gramme„ |
These differences in the water contents would cause differences of 0·00008, – 0·00002, and – 0·00005 respectively in determinations of the acid of 44·1 per cent, concentration (vide infra), which would be either diminished or increased by one or two units in the case of acids of 100 per cent. and 10 per cent. concentration respectively.
Correction for Displacement.—All the weighings for the determinations were reduced to a vacuum according to the usual approximate formula
W = ,!-- + -,
V P v)
wherein A. = 0'0012 is the adopted standard density of air under average conditions, and p taken as at 8 -4; the circumstance that the
smaller weights were of platinum instead of brass affects the results beyond the limits of experimental error and can therefore be neglected. The densities were throughout reduced in terms of the maximum density of water at 4 C.
Thirdly. Temperature.—The temperature was recorded by a thermometer, graduated to O·1 C., and of such an open scale that 0·03 C. could be read without difficulty; the zero displacement and the stem errors of the instrument were determined in our previous investigation.
With a view of ascertaining if there existed any appreciable error in the thermometer (by Casella) used, and to examine the accuracy of the method, the density of a sample of 44·10 per cent. acid (selected as approximately the mean of water and 100 per cent. acid), was
