176 



Estimation of the Specific Gravity of Blood. [Nov. 25 7 



The difference in the error of the several hydrometers is readily 

 accounted for. A consideration of the facts already set forth shows 

 that the error when expressed in scale units must vary directly as the 

 radius of the stem, and inversely as the total weight of the instrument, 

 so that when the stem is fine in comparison with the weight, the error 

 is small. But in making a small hydrometer it is impossible to keep 

 down the relative proportion of the stem. It thus follows that the 

 smaller hydrometers exhibit a greater surface tension error than the 

 larger ones. 



The greatest discrepancy^ which I have observed in any hydrometer in 

 chloroform and benzol was 0*014. This instrument was a small one, 

 graduated from 1*020 to 1*080, and is not included in the above 

 tables. 



This source of error in Hammerschlag's method may be obviated 

 by:- 



(1.) The estimation of the specific gravity of the chloroform and benzol- 

 mixture by means of an instrument which excludes or minimises the 

 surface tension factors. The most convenient is some such balance as 

 AVestphal's, in which the surface of the fluid is intersected by an 

 exceedingly fine platinum wire only. The employment of hydrostatic 

 bubbles is inconvenient on account of the long series required. The 

 picnometer method is not readily applicable in connection with very 

 volatile fluids. 



(2.) By employing a hydrometer which has been standardised or 

 corrected in chloroform and benzol mixtures, the requisite specific 

 gravities of which have been adjusted by an accurate method. In the 

 absence of the above-mentioned appliances a rough method or correc- 

 tion may be applied to any hydrometer of which the highest mark 

 is 1*000. The method consists of adjusting the proportions of a 

 mixture of chloroform and benzol until a small drop of water immersed 

 neither sinks or floats. The mixture being thus of the same specific 

 gravnVy as water itself, the reading of the hydrometer in it is its error 

 at this degree, and in the case of a well-constructed and accurately 

 graduated hydrometer, this error holds good with only a negligible 

 increase throughout the scale. 



I have, in conclusion, to express my indebtedness to Professor 

 Vaughan Harley for the resources of his laboratory, and to Professors 

 Baly and Donnan, of the Chemical Department, at University College., 

 for kind assistance and the loan of appliances. 



