It having been shown that dilution materially affects the separation of caffeine as periodide, the complete solution of the acetanilide-caffeine mixture was not attempted. Instead the residue consisting of acetanilide and caffeine, obtained by extracting with chloroform and distilling the solvent, was dissolved in hot water, the solution allowed to cool and, when the larger part of the acetanilide had again separated out, the liquid filtered, the residue washed with cold water and in the filtrate the caffeine precipitated and estimated as before outlined. Using about 1 gm. acetanilide and weighed portions of the caffeine solution, representing about .1 gm. caffeine, the caffeine so isolated represented: 98%, 98.7%, 97.4%, 98.4%, 99.1%, 97.9%; average 98.2%.

It was shown that the loss of nearly two percent of the caffeine was not because of the solubility off caffeine periodide in acetanilide 7 nor was evidence had that it was because of caffeine held back in the acetanilide which crystallized out.8

When the acetanilide-caffeine residue was dissolved in 50 c. c. hot water, filtered when cool, the flask and filter washed with 25 c. c.

7 When weighed portions of caffeine solution, representing about .1 gm. caffeine were diluted with 80 c. c. water, 1 c. c. hydrochloric acid added, then halfnormal iodine solution, 99.2 p. c., 99.7 p. c., 99.7 p. c., 99.4 p. c.: average 99.4 p. c. of the caffeine was recovered. When instead of 80 c. c. water. 80 c. c. of a saturated solution of acetanilide was added, then the results were 98.8 p. c., 99.3 p. c., 99.4 p. c., 99.4 p. c., 99.6 p. c., average 99.3 p. c.

8 When the acetanilide remaining in the funnel was washed with a little water. the washing remained clear when hydrochloric acid and iodine was added. When the acetanilide was dissolved in hot water, the solution cooled and filtered, no precipitation occured when hydrochloric acid and iodine was added,

water, 1 c. c. hydrochloric acid, U. S. added and then 25 c. c. fifthnormal iodine solution9 added somewhat better results were obtained. Thus, using 1 gm. acetanilide and weighed quantities of caffeine solution, representing about .1 gm. caffeine, then the caffeine so recovered represented: 99.3%, 99.1%, 98.3%, 98.5%, 98.7%, 99.0%, 98.5%, 98.2% or an average of 98.7%. Below are given a series of determinations of acetanilide and caffeine in which this proceedure was employed, thus: To the acetanilide contained in a separator, was added a weighed portion of the caffeine solution, 2 c. c. normal sulphuric acid 10 and sufficient water to make the total volume about 30 c. c. This is then extracted with five portions of chloroform, 20 c. c. each, and the extractions, passed through asbestos, received in a flask. The chloroform is distilled slowly from a water bath and when it ceases to pass over, the flask is detached, rotated until crystallization sets in and then either dried at 50-60 degrees during 12-24 hours (a) or else at 95-100 degrees during two hours (b) and weighed. Then 50 c. c. water are added to the contents of the flask and slowly heated to boiling. The clear solution is cooled and rotated occasionally until the excess of acetanilide separates out, adding a trace of acetanilide if crystallization does not set in promptly. The liquid is filtered through a small pledget of cotton into an Erlenmeyer flask and the transfer completed with 25 c. c. water. Then 1 c. c. hydrochloric acid U. S. P. and 25 c. c. approximately fifthnormal iodine solution, 25 gm. iodine and 36 gm. potassium iodide in 1000 c. c., are added and the mixture rotated occasionally during one-half hour. The clear supernatant liquid is decanted into a small short stemmed funnel containing a looselypacked plug of asbestos; finally the precipitate is transferred and the flask and precipitate washed with 25 c. c. tenthnormal iodine solution to which has been added .25 c. c., or 5 drops, hydrochloric acid. When the wash fluid has passed, the funnel is placed in the neck of a separator, about 2 gm. sodium sulphite introduced into

9 Attempts to decrease the total volume of the liquid in which the periodide precipitation is made, suggested the use of a more concentrated iodine solution. It was found that a clear solution resulted when to 90 c. c. of a saturated solution of acetanilide was acidulated with 1 c. c. hydrochloric acid and then 10 c. c. of an approximately halfnormal iodine solution, 63 gm. iodine and 90 gm. potassium iodide in 1000 c. c. was added. But, if either the iodine solution or the hydrochloric acid was materially increas d then precipitation took place. When a fifthnormal iodine solution was used no precipitation occured even when it was added in large quantities and in presence of much acid.

10 In acetanilid mixture containing sodium bicarbonate enough acid must be added so that the liquid is strongly acid toward methyl-orange T. S.

the flask in which the precipitation was made, 10 c. c. water added, rotated to dissolve any adhering caffeine periodide, the solution poured into the funnel containing the precipitate with a short glass rod.11 When the liquid has passed, the flask and funnel are rinsed with several small portions of water, using about 15 c. c. in all. The liquid is now extracted with five portions, 10 c. c. each, of chloroform, the extractions are filtered through asbestos into a small flask, 12 the solvent distilled slowly from a water bath in such a way that active ebullition does not occur and the flask dried at 95-100 degrees during three hours.

The weight of acetanilide was obtained by subtracting the weight of caffeine so found from the total weight of the chloroform extraction obtained (a. or b.).

13 The determinations marked (b) were made by drying the chloroform extract containing the acetanilide plus caffeine at 55 degrees, the other by drying at 95 degrees during two hours.

11 The amount of sodium sulphite is largely in excess of that required to dissolve the periodide and yields a colorless alkaline solution. If, because of deterioration of the sodium sulphite, the filtrate is yellow and shows an acid reaction when a drop of litmus, or cochineal T. S. is added, then it may be rendered alkaline with ammonium hydroxide.

12 If close results are desired, a flask of nearly the same size and weight should be used as a counterpoise.

The Balance.

By I. W. Brandel and Edward Kremers.

Historical Introduction. With the transition from the agricultural to the commercial stage of civilization, the comparison of masses early became an important operation. The simplest means of effecting such a comparison was found in the balance (bilanx, bilancis, having two scales; from bis, twice, and lanx, plate, scale). The earliest references to this instrument are possibly found in Egyptian papyri.

MADN

Fig. 1. Trade in Silphion, which was used for medicinal and culinary purposes. From Baumeister, Denkmäler des klassischen Alterthums,

Bd. III. p. 1664, fig. 1729.

The simplest form of the balance consisted of a beam, supported in the middle and later suspended at the same point by a ring or chain. From the ends of the beam the scale-pans were hung. The

rising or falling of the scale-pans indicated the greater or lesser weight. Illustrations which show this simple instrument are found e. g. on an Egyptian papyrus also on a Greek vase (Fig. 1). In the former instance, the beam is supported by a column in the latter it is suspended from the spat of a ship.

The first improvement of importance consisted of an index fastened at right angles to the beam. It enabled more accurate weighing and was known to the later Greeks and to the Romans. The latter called this index, pointer or tongue examen (for exagmen, from exigere, to weigh accurately); and the holder between which it moved, the vagina, because of its resemblance to the scabbard or sheath. Possibly the earliest Greek occurence of examen and vagina is found on the Kairos relief which, however, is very imperfect and indistinct. From the Roman period actual balances with these attachments are still in existence.

Fig, 2. Pompeyan Balance.

From Baumeister, Denkmäler des klassischen Altertums. Bd. III, p. 2078, fig. 2316.

A further step in the evolution of the balance is shown in the illustration of a Pompeyan balance (Fig. 2). One arm of the beam is graduated and provided with a rider as an efficient substitute for weights.

During the Roman period the equal arm balance seems to have been used primarily for larger operations (comp. Fig. 1), whereas for the weighing of small amounts an unequal arm balance, the desemer or statera apparently was preferred. This same instrument is still largely in use in Italy and other southern European

countries.