DAIRY FAEMING^ — DAIRYING. 675 



The total solids were determined " by drying from 1 to 2i gm. of milk in 

 dishes from 4 to 10 cm. diameter, some platinum, some porcelain, some dried in 

 an air-oven, some in a steam oven, some quiclily, some slovrly. It -was found 

 that the probable variation of a single determination from the mean is ±0.068. 

 This figure represents the probable variation from the mean, not the probable 

 variation from a superior method, as in the case of the lactometers. 



" Where the conditions are rigidly uniform, the probable variation from the 

 mean will be less than this figure ; but when results from difiiereut laboratories 

 are compared, I think that ±0.068 per cent total solids reasonably represents 

 the variation of a single determination in one laboratory from the mean of 

 many results in many laboratories. Where laboratories are not well equipped 

 the error may be greater." 



The fat was determined by the use of the Gerber method. Without any 

 attempt at standardization the probable variation of a single determination 

 from the mean of all determinations was ±0.036 per cent. The methods of 

 calculating the probable error in determining solids-not-fat are given and 

 show that these calculations produce the important result that the small lacto- 

 meters often sold are quite useless for determining the solids-not-fat, but that 

 if a lactometer with a fairly open scale be used quite good results can be 

 obtained. 



The results of testing the morning milk of a herd of 5 cows at Offerton Hall 

 for 5 years are given in the form of a frequency curve, showing that the distri- 

 bution of fat percentages follows quite closely the normal probable curve. The 

 use of the slide rule is recommended in applying the theory of errors. 



Other conclusions reached are the following: "The error of sampling small 

 quantities of milk by experienced men is negligible, but the error of sampling 

 large quantities of milk by inexperienced men is so great that the analysis is a 

 mere farce. The error due to the milker is a personal matter, impossible to 

 express in general terms. The error due to the times of milking is a subject on 

 which much work has incidentally been done. By compiling many statistics on 

 the subject, I calculate that if cows are milked 24 minutes later in the morning 

 than they used to be, no other change taking place, the fat will, on the average, 

 be 0.1 per cent lower than it used to be in the morning. Hence, irregularity in 

 times of milking will cause variations. 



"At Offerton Hall, with the experimental herds of 5 cows, the average differ- 

 ence between to-day's test and yesterday's test is 0.22 per cent fat. It is quite 

 impossible to say whether this figure is representative of any general state of 

 affairs. It is quite clear that the errors of experiment are greater than the 

 errors of analysis, and it should be obvious that the fluctuations of practical 

 dairying are greater than the errors of experiment. 



" The error of the solids-not-fat depends upon the method used, but by the 

 use of the Gerber machine, a good lactometer and a slide rule, the error can be 

 reduced so that 8.45 per cent solids-not-fat can be read as S.45±0.05; that is, 

 that the truth probably lies between the limits 8.40 and 8.50." 



Roumanian sheep cheese, D. Frangopol (Milchw. ZcntU., 7 (1911), No. 7, 

 pp. 289-296). — The methods of making several varieties of cheese are described 

 and analyses are reported. 



The cause of the dark coloration on the surface of Harz cheese, A. Wolff 

 (Milchw. ZentU., 7 (1911), No. 7, pp- 296-303) .—The organisms present on the 

 surface of the cheese were isolated, and the cause of the defect was thought 

 to be due to the action of several organisms, the chief one being a species of 

 bacteria to which the name of Bacterium denigrans was given. Subsidiary 

 causes were i)eptonizing bacteria, Mycoderma casei, and other organisms. 



