1896.J MICROSCOPICAL JOURNAL. 173 



cent depending upon the intensity of the lif^ht. The 

 Fleischl instrument is liable to an error of about ten per 

 cent. In the specific-gravity method there is liability of 

 error from two sources. The drop of blood may adhere 

 to the yides of the jar, or some air may become mixed 

 with it. These errors in the specific-gravity method are 

 reduced to a minimum by careful manipulation. 



The greatest error in this last method may be due to 

 the table, since of the cases from which liammerschlag 

 constructed his table, a great number were primary 

 anajmias and chloroses. For these his table would prob- 

 ably be more accurate than for our cases, as all the 

 ansBmias which we examined were secondary. Our cases 

 were taken as ordinarily found in hospital wards, both 

 medical and surgical, and covered a wide range of dis- 

 eases. 



We are convinced from the experience of others and 

 from our own observations that all of these methods are 

 liable to considerable error. Osier says that the error in 

 the Fleischl instrument may not be more than two per 

 cent in blood, which is nearly normal, but cites Neubert 

 and Letzius as having shown that in a much impover- 

 ished blood the error may be as much as twenty per 

 cent. 



The specific-gravity method has the advantage of 

 cheapness and convenience ; of taking but little blood, 

 and of not being a color test. This last is of the most 

 importance since the accuracy of the test does not depend 

 so much upon the judgment of the individual, and makes 

 it practical for observers who lack sufficient appreciation 

 of colors and shades. 



In following up a case with a color test, an error of 

 five per cent too low might be made at the first reading, 

 and one of five per cent too high at the second and the pa- 

 tient be supposed to have improved to that extent, when, 

 in reality, his condition had remained unaltered. With 



