812 EXPERIMENT STATION RECORD. 



iiut iu all cases show whether the milk has beeu skimmed, nor whether skim 

 milk has been added. 



Test of milk by means of carmin indigo, Vaudin {Rcr. Hijg. et Pol. Sanit., 

 29 (1907), A^o. 12, pi). 106.'>-106'J). — In making the test the sample of milk con- 

 taining a small quantity of the color solution is placed in a flask, tightly corked, 

 and kept in diffused light. The color gradually disappears, the time depending 

 on the initial germ content. Millv drawn with no care for cleanliness destroys 

 the color in a few hours, but when drawn under proper sanitary conditions it 

 does not lose color after 24 hours or more at about 60° F. Directions for pro- 

 ducing the coloring solution and measuring its strength are given. 



Determinations of the w^ater content of butter (Ber. K. Yet. og. Landbo- 

 JifJjsJcolcs Lab. Landtikonom. For.sng [CojicnlKiffoi], 6'.? {1901), pj). 30, figK. .'/). — 

 Butter Avas prepared for testing by melting a large representative sample and 

 shaking it violently while it cooled. Tlie water content as determined by the 

 ordinary laboratory metliod of chemical analysis (drying .5 gm. in a small por- 

 celain dish with about 15 gm. granulated pumice stone at 100° for 2 hours) 

 averaged 14.G6 per cent, whereas by the English metliod (drying 10 gm. over 

 direct Hame in a porcelain dish for about 10 minutes, using a glass rod for 

 stirring) it averaged 14.79 per cent. In only 6 out of 50 comparative determi- 

 nations were higher results obtained by the laboratory method. Results by 

 tlie Gray method were in all cases lower than those by the laboratory method, 

 averaging 0.31 per cent. The Englisli method is recommended for work in 

 creameries, and a special scale constructed for this purpose is described and 

 illustrated. A table showing the proportion of water in butter corresponding 

 to ranges in the weight of the fresh samples from 9 to 10.15 gm., and of the 

 dried samples from 7.15 to 9.1 gm., is given. 



Determination of the fat of cheese by means of carbon tetrachlorid, C. 

 MoNTANARi {Htdz. Spcr. Aqi'. Itah, .',0 {1901), No. 2, pp. i7_/-/7' 6"). —Results 

 obtained by the use of carbon tetrachlorid for the determination of the fat of 

 cheese are given, in comparison with those br the use of ether. Utilization of 

 the former reagent is recommended. 



Casein testing {Paper Making, 26 (1901), No. 12, pp. 5Jift-546). — The users 

 of casein in manufacturing are recommended to buy only a pure product, and 

 to dissolve it themselves, in preference to using so-called "soluble" casein. Di- 

 rections are gi\en in some detail for testing the casein. 



The analysis and grading of creosotes, A. I. Dean and E. Bateman {U. >s'. 

 Dept. Agr., Forest Scrv. Circ. 112, pp. J/'i, figs. 28). — Investigations reported in 

 a previous publication (E. S. R., IS, p. IIOS) have been continued. Consider- 

 able data have been gathered regarding the fomposition of creosotes and in 

 the present publication data are reported regarding the volatility of creosotes, 

 properties of pure coal-tar creosote and oils other than coal-tar creosote, and 

 of mixtures of coal-tar creosote and other oils. The value of thorough creosote 

 analyses is considered and improvements in the method of distillation used in 

 the analyses are described. 



In the authors' opinion it seems desirable that there should be two methods 

 of analyzing creosote — a field method to show the general character of the oil 

 imder analysis, and a laborati>ry method adai)ted to the analysis of oil which 

 for any reason requires careful examination. The details of such methods are 

 given. Various grades of creosote and their uses are explained. 



Regarding the new international method of tannin analysis, K. Schor- 

 LEMMER and H. Sichlino {CoUcginni, 1907, No. 2H8, pp. .'f35-4'i0). — Results ob- 

 tained in the analysis of various tannin substances by the present method 

 (E. S. R., 19, p. .309) and by the old method are compared. The authors be- 

 lieve that the data show that the newer method gives the more correct figures. 



