MICROSCOPICAL EXAMINATION OF FLOUR. 



9 



The variation in the counts on these slides naturally raises the ques- 

 tion of the limits of accuracy in weighing out the test portion of flour. 

 Since the amount of flour used on a slide is 5 milligrams, it is desirable 

 to determine how great is the error due to weighing the test portion of 

 flour. The balance employed in this investigation was a fine assay 

 balance. In weighing the sample the vibration method was used, 

 and the quantity of flour was so adjusted as to produce a deviation 

 of approximately not more than one-fourth of a space on each side of 

 the zero point of the scale. This is equivalent to not more than 1/40 

 milligram, or one-half of 1 per cent, on the basis of the portion of flour 

 used (5 milligrams) . Hence any error in weighing can not be accepted 

 as an explanation of the difference in slide counts. 



VARIATION IN HOMOGENEITY OF BULK SAMPLE. 



The question has been raised as to whether or not a portion of the 

 slide variation might not be accredited to lack of uniformity of the 

 bulk sample, due to the fact that any grade of flour is usually the 

 component result of several constituent streams which vary more or 

 less among themselves. The fact that in general practice the flour 

 stocks are subjected to a certain degree of purification, however, 

 leaves this factor little chance to figure to any great extent. This 

 point was tested by passing a certain sample of flour which had an 

 average count of 32 bran particles and 64 hairs through a 30-mesh 

 sieve and making up and counting 12 slides. The bulk sample was 

 then passed through the sieve once more (making two times for the 

 sample), and another series of slides made and counted. Finally, 

 the sample was put through the sieve twice more (making four times 

 for the sample) , and a third series of 12 slides made and counted. The 

 results of these tests are given in Table 6, the counts in which are the 

 average of the results obtained by two persons. 



Table 6. — Effect of variation in homogeneity of sample on count. 



Sample passed through 30-mesh sieve- 



Once. 



Twice. 



Four times. 



Bran 

 particles. 



Hairs. 



Bran 

 particles. 



Hairs. 



Bran 

 particles. 



Hairs. 



31 

 48 

 36 

 33 

 41 

 36 

 35 

 35 

 37 

 35 

 30 

 34 



73 

 73 

 58 

 53 

 64 

 64 

 74 

 75 

 57 

 76 

 66 

 60 



26 

 37 

 25 

 22 

 27 

 32 

 32 

 27 

 34 

 26 

 39 

 30 



67 

 70 

 58 

 54 

 61 

 57 

 83 

 64 

 61 

 72 

 67 

 67 



21 

 22 

 30 

 32 

 23 

 29 

 36 

 34 

 39 

 38 

 33 

 37 



76 

 53 

 56 

 81 

 79 

 66 

 60 

 56 

 70 

 52 

 65 

 48 



135 



218 



166 

 2 23 



129 

 217 



165 

 2 29 



131 



U8 



163 



2 31 



1 Average. 

 152332°— 20— Bull. 839- 



1 Variation 



