Nuttall and Stningeivays 313 



such accidents. We know that the" concentration of the blood may vary 

 in disease and even in health, and it is possible that blood remaining 

 in the blood vessels after death may undergo changes in concentra- 

 tion. Apart from this there are chemical changes taking place in the 

 blood of which we have but slight knowledge at present, which may 

 materially affect the quantitative results of the precipitin test. The 

 tests were conducted upon bloods of very different ages, some preserved 

 with chloroform, others filtered through porcelain and preserved pure. 

 In most cases serum alone was used, in some instances clots were 

 present in the serum. All the bloods were kept in the ice-chest until 

 used. In a few cases putrefaction had already taken place when we 

 received the bloods. The bloods were kept in tightly stoppered 

 bottles, or in sealed tubes. In warm weather it is reasonable to 

 suppose that a certain amount of evaporation may have taken place 

 from the bloods which were allowed to clot in the covered vessels in 

 which they were placed for obtaining serum. Where glass stoppers 

 were used a small amount of vaseline was smeared upon the ground- 

 glass surfaces. In the bloods first collected a slight loss of water by 

 evaporation may well have taken place through the use of corks which 

 loosened or were defective. All bloods of later date were kept in sealed 

 tubes to prevent evaporation. 



The quantitative method devised by Nuttall is also open to criticism. 

 It appears a priori improbable that the physical character of the pre- 

 cipitum will be the same in different bloods to dilutions of which a 

 given antiserum is added. We may assume the precipitum should 

 " pack " to a greater extent in tubes containing the most and to various 

 degrees in tubes of inconstant calibre. Finally, there is considerable 

 chance of error in the measurement of such small quantities as have at 

 times to be dealt with. A considerable number of bloods were tested to 

 determine the range of experimental error, the result being that we 

 found the error to be within 10 "/o- Thus Graham-Smith and Sanger 

 (1903, p. 264) obtained the following results : 



^^ Results of measurements of fo\ir samples of a 1 in 21 dilution of 

 human blood and 8 samples of a similar dilution of ox serum. 



Meau -0233 c.c. 



