of corrections for (i) percentage proportion of nitrogen peroxide, and (ii) variations of temperature.
These writers further pointed out that if the values for the densities of acids above 70 per cent, be represented in terms of percentages, such values lie upon a curve and not a straight line, as indicated by the determinations of Kolb. Various isolated determinations of acids of various strength have been published by Kohlrausch and Perkin for special purposes in the course of investigations on electric conductivity and specific magnetic rotatory power.
V. The Sources of Error in Pyknometers.
The sources of error may be classified under three categories, namely, (i) weight of glass and its contents taken by themselves ; (ii) tempera- ture ; and (iii) adjustment.
Firstly. Weight of Glass and Contents. These arise from (i) corrosion, .and (ii) absorption.
To avoid the second error of absorption the acids were taken in either a gradually ascending or descending series of concentration ; between any set of observations the tube was not washed out with water, but rinsed frequently with the acid to be subsequently used. The object of this 'precaution is to gradually season the vessel, as it were, for the particular acid to be examined, as undoubtedly some of the acid is soaked into the glass ; if, then, in a subsequent observation the acid formerly absorbed is soaked out and replaced by an acid of different concentration, the error produced in the weight of glass and acid is minimised. The methods recommended by various writers of (Cleansing with water, steam, or dry air were found to cause error, while the use of alcohol and ether, formerly in vogue, was quite iinadmissible.
Secondly. Adjustment. The instruments were placed in a water- bath at the required temperature, and the solutions drawn off from one limb by bibulous paper until the level-mark on the opposite limb was .attained ; for the most concentrated acids, which at once nitrate and render useless the common form of bibulous paper, the only material iound suitable was a kind of tissue paper, sold for a wholly different purpose.
In order to estimate the effect of an error purposely committed in ihe adjustment, the following method was adopted : One of the . Pyknometers filled with water was placed in a water tank, and a horizontal microscope with vertical movement was so arranged that the adjustment mark was in the field of view ; on the eyepiece of the microscope was etched a 1 cm. scale, divided into hundredths, so that a difference of a deci-millimelre could be read off. As the adjustment mark appeared too coarse an object under the magnification, a brilliant
