302 



DR. ROBERTS ON THE ESTIMATION 



TABLE I. 



No. 



Sugar per 100 parts 



by the formula 

 S = (D-D JxO-23. 



Sugar per 100 parts 



by direct volumetrical 



analysis. 



Difference. 



I 



*^ 



3 



7-51 



7-47 

 6-68 



7-69 

 7-69 

 6-66 



o-i8 



0'22 

 0"02 



*4 



6-72 



666 



o"o6 



5 



5-i6 



5-i8 



0'02 



*6 



5-19 



5-i8 



o-oi 



7 

 #8 



5-65 

 5-65 



5-77 



5-77 



0-I2 

 0-I2 



9 



4'47 



4-35 



0-I2 



*IO 



II 



4"49 

 7-85 



4-35 

 8-o6 



0-14 

 0-21 



12 



5-91 



6'io 



019 



13 



11-27 



ii'36 



0*09 



*i4 

 16 



II-3I 



5'69 

 811 



11*36 

 5-68 

 8-o6 



0-I5 

 o-oi 



0-05 



*i7 



8-09 



8-o6 



0-03 



*i8 



8-00 



8-o6 



o'o6 



19 



8-29 



8-20 



0-09 



20 



7-61 



7-74 



0-13 



These results are so close that the discrepancies may 

 be regarded as due to errors of manipulation rather than 

 to any fault in the method. 



In pursuing the inquiry further it was found that the 

 volumetrical analysis, in spite of every care in its perform- 

 ance, did not possess the delicacy and certainty required 

 in a standard when minute differences were concerned; 

 insomuch that, when a discrepancy appeared between the 

 indications of the volumetrical and the fermentation me- 

 thods, it was found impossible to decide in which pro- 

 ceeding the error lay. 



In order therefore still further to test the constancy of 

 the results, artificial diabetic urines were prepared by 

 diluting a natural diabetic urine with known volumes of 

 water, or of a healthy non-saccharine urine. Assuming the 

 estimate of the sugar in the original urine to be correct, 

 the quantity of sugar in the dilutions could be ascertained 



* Those marked with an asterisk are duplicate experiments. 



