172 THE AMERICAN MONTHLY [May 



cent. We have also found that in one-fiftli of our cases 

 we disagree in our readings of the same instrument. 



We have found it a great inconvenience in making bed- 

 side tests in a hospital ward, to run to some other part 

 of the ward or building (to a dark room). In order to 

 obviate this difficulty we have adopted the following 

 device : This consists in our instrument bag fitted witli 

 a cardboard cover ; at one end of tliis a hole is cut for the 

 passage of a lamp chimney ; at the otlier end a small hole 

 for looking through the well of the instrument, and at 

 one side of this a window with a flap for inserting the 

 hand to move the wedge. 



Hammerschlag's method has the advantage that there 

 is no color test. Every one must agree as to wliether 

 the drop rises or sinks or stays where placed. It is also 

 very inexpensive, all that is necessary being a hydrometer 

 jar, chloroform and benzole. The method of Roy and 

 Jones necessitates keeping on hand a large number of 

 solutions which require careful standardization and must 

 be re-standardized at frequent intervals. Although this 

 method may be better where a large number of cases are 

 to be examined in a short time, yet for the ordinary ob- 

 server who uses a method of this kind less often and upon 

 a small number of cases, the one which we have used 

 seams preferable. 



In both methods, Hammerschlag and Jones have found 

 that there is no appreciable difference due to variations 

 of temperature in the room. 



The results which we have obtained in making })arallel 

 tests with the above described methods, may be sum- 

 marized as follows : 



The readings of the Fleischl ran as a rule from ten to 

 fifteen per cent lower than the percentage estimated 

 from the specific gravity. The readings of the Gowers 

 ran a few per cent lower than the specific. The Gowers' 

 instrument is liable to an error of at least fifteen per 



