In the following paragraphs are presented the methods adopted for the estimation of the various constituents of teas.

Caffeine or theine, C8H10N4O2. - After experiments with a number of methods for the estimation of caffeine, the following was adopted on account of its simplicity and. the accuracy of the results:

1Transfer 3 gs of powered tea into a 300 cc flask, add about 250 cc of water; heat gradually to the boiling point, using a small fragment of tallow to prevent frothing; boil gently 30 minutes, washing down any particles of tea which may collect on the flask above the liquid. After the liquid begins to boil the flask should be filled almost to the neck and water should bo occasionally added to keep its level at this point. Several samples may easily be extracted in this way at the same time, since they require but little attention after regulating the flame of the lamp. After boiling 30 minutes, cool, add a strong solution of basic acetate of lead in sufficient quantity for the removal of precipitable substances; usually about 3 cc are required; complete the volume to 300 cc, mix thoroughly, and filter, rejecting the precipitate unwashed. Treat 50 cc of the filtrate with H2S for the removal of the lead; boil off the excess of H2S; filter into a separatory funnel, washing the precipitate with hot water, or an aliquot part of the filtrate may be taken and washing of the precipitate may thus be avoided. The error from the volume of the precipitate is very slight.

Extract the water solution in the separatory fuunel seven times with chloroform; collect the chloroform solution of caffeine in a small tared flask, and remove the solvent by distillation; dry the caffeine at 75° C. two hours, weigh, and calculate the percentage.

1 Jour. anal. Chem. 4, 4, p. 390.

Nitrogen. - Tea contains a very high proportion of nitrogen. The analyses of Mr. Kozai, tabulated on page 879, will show how this nitrogen is distributed in the prepared leaf.

The nitrogen of teas should be determined by the absolute method, or Kjeldahl method, modified for alkaloids. These methods are fully described in the text-books on quantitative analysis, and in Bulletin 24, page 217, of the Chemical Division of this Department.

Albuminoid nitrogen. - Stutzer's1 method. - Prepare cupric hydrate as follows: Dissolve 100 grams of pure cupric sulphate in 5 liters of water, and add 2.5 cubic centimeters of glycerin; add dilute solution of sodium hydrate until the liquid is alka-liue; filter; rub the precipitate up with water containing 5 cubic centimeters of glycerin per liter, and then wash by decantation or filtration until the washings are no longer alkaline. Rub the precipitate up again in a mortar with water containing 10 per cent of glycerin, thus preparing a uniform gelatinous mass that can be measured out with a pipette. Determine the quantity of cupric hydrate per cubic centimeter of this mixture. To 1 gram of this substance add 100 cubic centimeters of water in a beaker; heat to boiling, or, in the case of substances rich in starch, heat on the water bath ten minutes; add a quantity of cupric hydrate mixture containing 0.7 to 0.8 grams of the hydrate; stir thoroughly; filter when cold; wash with cold water, and put the filter and its contents into the concentrated sulphuric acid for the determination of nitrogen after Kjeldahl. For the above filtration use Schleicher and Schiill's No. 589 paper, or Swedish paper, either of which contains so little nitrogen that it can be left out of account.

Tannin. - Lowenthal's2 method an improved by Councler and Shroeder, and Procter. - In the estimation of tannin by this method the following standard solutions and reagents are required:

(1) Potassium permanganate solution containing, approximately, 1.33 grams of the salt per liter.

(N / 10)

(2) Tenth-normal (N / 10) oxalic, acid solution, tor use in standardizing the permanganate solution.

(3) Indigo-carmine solution, containing 0 grains of this indicator and 50 cc concentrated sulphuric acid per liter.

(4) Gelatin solution, prepared by swelling 25 grams of gelatin for one hour in a saturated solution of common salt, then heating until the solution is complete, and finally, after cooling, milking up to one liter (W. H. Krug's method of preparing this solution).

(5) Salt-acid solution, prepared by mixing 975 cc saturated common salt solution and 25 cc concentrated sulphuric acid.

(6) Powdered kaolin.

The potassium permanganate solution is standardized in the usual manner by oxalic acid. It is obvious, in the analytical manipulations which follow, that the end reaction (golden yellow or pink tinge) which is adopted in the first process must also be employed in the second. The indigo-carmine should be very pure and especially free from indigo-blue.

(a) Five grams of finely powdered tea are placed in a flask of approximately 500 cc, capacity and boiled thirty minutes with 400 cc distilled water. The water should be cold when added to the sample. After the completion of the boiling, the flask is cooled and the solution and residue are transferred to a half-liter Mask and the vol-ume made up to 500 cc. The solution required for analysis is filtered off if neces-sary. To 10 cc of the tea infusion, 25 cc indigo-carmine solution are added, and approximately 750 cc distilled water. The permanganate solution is now added, a cubic centimeter at a time, the liquid being vigorously stirred after each addition, until the color changes to a light green; the addition of permanganate is continued more slowly, drop by drop, until the whole liquid takes on a bright golden-yellow color (Councler and Schroeder), or, if preferred, until the pure yellow liquid shows a faint pinkish rim (Procter). The burette reading is now taken, and furnishes the value (a) of the formula. It is absolutely necessary to vigorously stir the liquid during the whole operation. It is best to repeat this titration, as well as that which follows in the next step of the analysis, and take a mean of several readings. (b) 100 cc of the tea-infusion (filtered if not sufficiently clear after decantation) are mixed with 50 cc of the gelatine solution in an Erlenmeyer flask, then 100 cc of the salt-acid solution and 10 grams of kaolin are added, and the whole vigorously shaken in the well-corked flask. Several minutes' shaking is necessary. If these directions are carefully followed the precipitate will settle very rapidly, leaving a clear, supernatent liquid which filters with great ease. The use of kaolin, as recommended by Procter, is an important modification of the original method, without which it is often impossible to separate the precipitate. The whole liquid is filtered and 25 cc of the filtrate (=10 cc of the original infusion) are mixed with 25 cc of indigo-carmine solution and about 750 cc water, and a titration made as under (a). The burette reading gives the value b of the formula.

1 Bulletin 31, Div. Chem., U. S. Dept, Agric, p. 189.

2Councler and Schroeder, Ztsch. anal. Chem. 25, 121. Procter, .Journ. Soc. Chem. Ind., 3, 82.

The value a is the amount of permanganate solution necessary to oxidize all oxidi-zable substances present; b, the amount required to oxidize the substances other than tannin; hence a - b = c = permanganate solution required by the tannin. According to Neubauer, .04157 grams of gallo-tannic acid is equivalent to .063 grams oxalic acid; therefore, knowing the amount of oxalic acid equivalent to the permanganate required to oxidize the tannin, we can easily calculate the amount of this latter substance present.

The above method for tannin was selected after experiments with several other methods. The general advocacy of its use by a number of very able analysts and the satisfactory results obtained in this Department led to its adoption for this work. Unfortunately, owing to the great variations in the results obtained by different methods and the acknowledged inaccuracy of many of these methods, the value of a large number of tannin determinations, by various authorities, is doubtful. For work of this kind the method of analysis should be clearly stated, and as far as possible, for the sake of uniformity, analysts should all adopt the same method in order that their work may be comparable with that of others.

Water. - The moisture may be determined in the usual manner by drying 1 or 2 grams of the powdered tea three hours at 100° C. in a flat dish, and calculating the water from the loss in weight. This method probably entails a slight loss of theine, which is credited to the moisture, but the error is very small and is negligeable.

Ash, total. - Two grams of the powdered tea are incinerated, at as low a temperature as practicable, and the percentage calculated as usual.

Ash, soluble and insoluble. - The total ash is treated on a filter with hot water until. all the soluble matter is dissolved; the solution is evaporated to dryness, and the residue ignited at a moderate temperature. The per cent soluble ash is calculated from the weight of this residue; the insoluble ash is determined by difference.

Ash insoluble in acid. - The water insoluble residue from the soluble ash determination is treated with hydrochloric acid. The undissolved portion is washed with water, dried and weighed, and its percentage calculated.

Ash, alkalinity. - The soluble matter obtained in determining the soluble ash is disN solved in a little water and titrated with acid. The alkalinity is calculated as potassic oxide (K20.).

Extract, total, and insoluble leaf. - Two grams of finely powdered tea are successively extracted with seven portions of 50 cc each of boiling water. The extract is decanted each time and the fractions united. The decanted solution is now boiled and passed through a tared filter. The insoluble residue is finally transferred to this filter, thoroughly washed with boiling water, dried, and weighed. The total weight, less the tare of the filter, is the insoluble residue from which the per cent insoluble leaf is calculated. The total extract is determined by difference, deducting the per cent of insoluble leaf from 100.

A weighing tube should be used in weighing filters and residues on filters.

Half-hour extracts. - This extract is of doubtful value for comparative purposes unless certain conditions are adopted and strictly adhered to. Slight variations may cause very perceptible errors in the results. I have followed essentially the conditions indicated by Geisler, which are evidently based on Wanklyn's method.

Place 1 gram of leaf tea in a 300 cc flask, add 100 cc cold distilled water and a small fragment of paraffin, to prevent foaming. Immerse the flask a minute or two in water heated to 90°C. in order to raise the temperature of its contents quickly to within a few degrees of the boiling point. Having dried the outside of the flask, boil the mixture thirty minutes. The flask must be fitted with a reflux condenser. After boiling, cool quickly in a stream of water and filter off the extract. Transfer an aliquot part of the extract to a tarred dish and dry. Calculate the per cent of matter from the weight of the residue.

General Remarks To Analysts

The analyst must be guided almost entirely by comparisons of his work on the suspected teas with the records of analyses of pure samples. The microscopic are almost the sole methods of detecting many of the adulterants of teas. Questions in regard to quality, where this depends upon aroma only, must be answered by a professional tea-taster.