1) Report of Roure-Bertrand Fils April 1912, 3. The authors recommend their method for the direct determination of geraniol. As a matter of fact all of the alcohols contained in the citronella oil are included in the estimation. Hence the determination of the so-called total geraniol enables the analyst to inform himself with regard to the citronellal content.

2) Hydroxylamine solution: 10 g. of hydroxylamine hydrochloride and 12 g. of potassium carbonate are dissolved each in 25 g. of water, the solutions are mixed and filtered.

3) Report of Schimmel & Co. October 1912, 43.

4) Americ. Perfumer 6 (1912), 284; Report of Schimmel & Co. April 1912, 75.

Schimmel & Co. carry out the method somewhat differently: 1 g. of Ceylon oil or 0,5 g. of Java oil (weighed exactly) are mixed with 10 cc. of freshly prepared 2 p. c. alcoholic phenyl-hydrazine solution and set aside for 1 hour1) in a 50 cc. flask provided with a glass stopper. 20 cc. of tenth-normal hydrogen chloride are then added and the mixture gently shaken. 10 cc. of benzene2) having been added, the mixture is shaken thoroughly and transferred to a separating funnel. After having stood for a short time, the 30 cc. of acid liquid, which separate readily, are drawn off and filtered through a small filter.

20 cc. of this filtrate, to which 10 drops of ethyl orange (1:2000) have been added, are titrated with tenth-normal potassium hydroxide until a distinct yellow color is attained. From this result the amount of tenth-normal potassium hydroxide requisite for 30 cc. of filtrate is computed. In order to ascertain the strength of the phenylhydrazine solution, the same test is conducted on a blank. If for the actual experiment a cc. and for the blank b cc. of tenth normal potassium hydroxide are used, then the citronellal equivalent of the amount of oil used (5 grams) is equal to a - b cc. tenth-normal potassium hydroxide. Since 1 cc. of tenth-normal potassium hydroxide corresponds to 0,0154 g of citronellal, the percentage of citro-nellal in the oil results from the following formula.

') According to the experience which Schimmel &Co. have thus far had with this method, it seems desirable to prolong the period of contact to 11/2 hrs. Report of Schimmel & Co. October 1912, 40.

2) The reason for shaking out with benzene is to clarify the solution which has become turbid upon the addition of hydrochloric acid. This enables a better observation of the changes in color when titrating.

(a - b). l,54 / s

Kleber records a citronellal content of from 7 to 9 p. c. for Ceylon oil, Schimmel & Co. found 7,5 and 10 p. c.

The citronellal content of citronella oil can likewise be ascertained according to the method suggested by V. Boulez1). However, as pointed out by Schimmel & Co., this method is applicable, only when the aldehyde content is not too high. Hence it is applicable to the Ceylon oil only. In an Erlenmeyer flask 25 or 50 g. of oil are shaken with 100 or 200 g. of a bisulphite solution previously saturated with normal sulphite and the mixture set aside for 2 to 3 hours until all of the aldehyde has combined. 100 or 200 g. of water are then added, the flask connected with a reflux condenser, and the mixture heated in a water bath until a complete separation has taken place between the oily layer and the aldehyde sulphonate solution. The oil is separated by means of a separating funnel and weighed. The loss in weight represents the citronellal content of the original oil. By means of acetylation of the non-dissolved portion of the oil the alcohol content can be ascertained. This method is not applicable to Java oil since, in consequence of the high citronellal content, the amount of non-aldehyde oil dissolved in the bisulphite solution becomes so great as to yield results that are too high2).

In a Ceylon citronella oil Boulez found 18 p. c. of citronellal whereas Schimmel & Co. found only 6,0 and 7,0 p. c. in two oils. It is noteworthy that the results found by Dupont and Labaune, also by Boulez on the one hand, and those found by Schimmel & Co. on the other hand differ materially from each other.

1) Bull. Soc. chim. IV. 11 (1912), 915. Report of Schimmel & Co. October 1912, 44.

2) Report of Schimmel & Co. October 1912, 45.

Whether these differences are due to the method or are due to the differences in the Ceylon citronella oil cannot be decided until the methods have been applied more extensively.

The formylation method as applied to citronellal (vol. I, p. 580) does not yield satisfactory results3), hence cannot be recommended.

Although the total geraniol assay leads to the best valuation of citronella oil, its application on the part of the non-expert is too difficult, hence it cannot be universally applied in commerce. For this reason the solubility test, which has come to be known as "Schimmel's Test"2) has become common. This test can be applied readily without special skill and is useful as a rapid test that throws light on the approximate nature of the oil. According to this test the oil is to be soluble in 1 to 2 vols, of 80 p. c. alcohol at 20°. The solution should remain clear when the amount of alcohol is increased to 10 vols, or should at most become slightly opalescent, but never separate oil even after standing for several hous. While adding the alcohol, violent shaking should be avoided, for otherwise such oil as has not gone into solution may be so finely suspended that a separation takes place only after prolonged standing, thus interfering with the observation.

This test is directed primarily against adulteration with fatty oils or petroleum, both of which are insoluble in 80 p. c. alcohol. In the case of citronella oil, petroleum has been a favorite adulterant.

Citronella oil adulterated with fatty oil does not form a clear solution with 80 p. c. alcohol. Adulterated with petroleum, the oil mostly yields a clear solution with 1 to 2 parts of 80 p. c. alcohol, but this solution becomes turbid upon the addition of more solvent. Moreover, fatty oil is apt to settle to the bottom, whereas petroleum rises to the surface. The exact proof of such adulteration is described on pp. 613 and 614 of the first volume.