In the case of the diphenacetyl derivative, the agreement between the experimental and calculated molecular volume is very satisfactory, and the numbers found for the diacetyl and monophenacetyl compounds do not point to any "association" of their molecules. The calculated value in the case of the monoacetyl compound is in close agreement with the theoretical (experimental, 209·1; calculated, 209-8), but has not been included in the table, as our preparation was not pure.

CONCLUSIONS.

The results which we have obtained in the examination of these compounds are plotted in Fig. I (p. 1106) in the form of curves, showing the variation of specific rotation with temperature. It will be seen, in the first place, that the rotations of the acetyl and phenacetyl derivatives behave quite differently, the former having a positive, and the latter a negative, coefficient, but this does not necessarily indicate some radical difference of properties of the substituting groups. These compounds may rather be regarded as extreme members of a series in which the temperature-coefficient, at first positive, gradually decreases, and ultimately becomes negative. The coefficient for diethyl di-monochloroacetyltartrate is, like that of diethyl diacetyltartrate, positive, whilst that of diethyl di-dichloroacetyltartrate is positive, but only very small (Frankland and Patterson, loc. cit.), and probably that of the di-trichloroacetyl compound, if it could be prepared, would be found to be negative, like that of the phenacetyl derivative.

It is noticeable that the coefficient of the monoacetyltartrate is greater than that of the diacetyl compound, this applying also to the phenacetyl derivatives, for the coefficient of the di-substituted compound is, in this case, more negative than that of the mono-substituted derivative, and we have, therefore, in the monoacyl series, the same regularity as before, namely, a considerable positive coefficient changing gradually on passing through acetyl, monochloroacetyl, trichloroacetyl (which has almost reached insensitiveness, but whose coefficient is slightly positive), to phenacetyl, whose coefficient is negative and large.

VOL. LXXVII.

4 F

A comparison of the molecular rotations of the diacyl compounds shows, in the first place, a striking qualitative difference between the first four derivatives of diethyl tartrate and the last four in the table; a division into two classes thus seems to be indicated, as in the case of the glycerates already quoted. At the same time, however, the acetyl and phenacetyl radicles have a different quantitative effect, for the introduction of the two acetyl groups somewhat diminishes the positive rotation of diethyl tartrate, the two benzoyl* and the two toluyl groups exert a very great influence in the same direction, whilst the two phenacetyl groups cause a slight increase in the positive rotation.

With regard to the effect of temperature, it will be noticed that in this series, the phenacetyl compound † has a negative temperature-coefficient but those of the toluyl and acetyl derivatives are positive. In the monoacyl series, it might be said that the phenacetyl and toluyl radicles have a similar function; exactly the opposite may be concluded in this diacyl series.

It would appear, therefore, that in one series of similar compounds the phenacetyl and toluyl derivatives behave alike, but in another differently, and that the data given seem as yet to be capable of no generalisation. This, however, is not the case, for, in fact, a much more comprehensive regularity than is at first sight apparent can be shown to exist. This is noticed if we compare the effect produced by the introduction of one acyl group into the diethyl tartrate molecule with that produced when two such groups are introduced. In every case so far examined, the rotation has been increased when one acyl is substituted for the hydrogen of one hydroxyl group, but when the hydrogen of the other hydroxyl is replaced by another acyl, then the positive rotation of the monoacyl derivative is diminished. The extent of this diminution varies greatly with the nature of the substituting acyl radicle; thus, with phenacetyl, the diminution is not so great as was the first increase, with acetyl the decrease is a trifle greater than the first increase, but with benzoyl and toluyl the diminution is so great that the diacyl compounds have a very high lævorotation. This regularity is clearly exhibited by the diagram (Fig. II, p. 1109) in which the abscissæ are taken proportional to the molecular weights of the compounds. The lines in this table have

The benzoyl compounds have been introduced in the tables in order to show how they compare with the toluyl derivatives.

+ This applies also to the benzoyl compound between the temperature limits 20° and 100°.

The point for di-p-toluyltartrate falls outside the table (at -484° for molecular weight 448,) but the inclination of the line shown is correct.

FIG. II. Comparison of the Molecular Rotations at 20° of Diethyl

no direct meaning, they merely serve to connect the three points, but it is seen clearly by their resemblance-initial rise and subsequent fall —that these acyl groups exert a similar qualitative influence.

YORKSHIRE COLLEGE,

LEEDS.

C.-Estimation of Atmospheric Carbon Dioxide.
By JAMES WALKER.

WHEN atmospheric carbon dioxide is estimated by means of Pettenkofer's method in its ordinary form, the results are often irregular and almost invariably too high from absorption of carbon dioxide from expired air during the process of titration. Many modifications have been proposed for avoiding this and other sources of error. For an account of these and for valuable accurate original observations, reference may be made to the memoirs of Blochmann (Annalen, 1887, 237, 39) and of Letts and Blake (Proc. Royal Dublin Soc., 1900, 9, 107).

The following modification, which I have used for some time, can be worked without any special appliances, and gives results which have an accuracy of 0.1 part in 10,000 under ordinary circumstances, whilst permitting the analysis of air containing any quantity of carbon dioxide from 0 to 40 volumes in 10,000 without any alteration in the mode of working.

The solutions employed are decinormal hydrochloric acid which has been exactly standardised, and a clear solution of baryta, the strength of which relatively to the hydrochloric acid is accurately known. The baryta solution is most conveniently made about 0.02 normal, and is kept in a stock bottle with a 50 c.c. burette attached, as described in Ostwald's Physicochemical Measurements, pp. 88 and 250.

The bottle in which the sample is collected is a clean, dry Winchester quart, the capacity of which has been previously determined. This bottle may either be furnished with a rubber stopper and tubes as described below, or with its own ground glass stopper covered with a very thin film of a stiff grease. In the latter case, immediately before the determination is to take place, the glass stopper is rapidly exchanged in the open air for a rubber stopper through which pass two glass tubes, about 7 mm. in diameter. The longer tube reaches almost to the bottom of the bottle; the shorter tube ends internally flush with the stopper. Both tubes project externally about 2 inches, and are provided with stopcocks at slightly different levels so as to permit of convenient manipulation. There is permanently attached to the upper end of the longer tube a piece of rubber tubing 1 inch in

length which serves to connect it with the jet of the baryta burette. This jet is best constructed of barometer tubing very slightly tapered and cut off square at the end. Fifty c.c. of baryta solution, which will suffice for 40 vols. of carbon dioxide in 10,000 vols. of air, are slowly run in with both stopcocks open. The rubber is then detached from the jet and compressed by the fingers in order to force the baryta solution which may remain in the tube down below the level of the stopcock. Both stopcocks are then closed, and the bottle, which is allowed to lie on its side, is agitated from time to time.

While the absorption of the carbon dioxide is in progress, an asbestos filter is prepared. The best form to use is a Soxhlet cuprous oxide tube, which is fitted into a filtering flask of about 200 c.c. capacity by means of a rubber stopper. A very small quantity of asbestos well teased out is pressed down over the capillary portion of the tube, the filter pump is then turned on, and several fills of distilled water run through the tube. When the pump is running at full speed, the water should flow from the filter tube in a continuous stream, but should only drop slowly when the suction is slight. After a little practice has been acquired, the preparation of a satisfactory filter only occupies a few minutes.

If the bottle has been repeatedly agitated, the absorption of the carbon dioxide may be regarded as complete in 15 minutes, and the filtration may take place. Ten c.c. of hydrochloric acid solution are introduced into the empty filtering flask by means of a pipette which is always used for the same purpose; the filter tube is fitted in, and the flask, resting on the sole of a retort stand, is clamped firmly by the neck. The bottle is now clamped in an inverted position on the same retort stand about 8 inches above the top of the filtering tube. Into this there is fitted a rubber stopper, through which passes a short glass tube, connection with the shorter tube of the bottle being completed by means of a piece of rubber tubing inch in bore,

in diameter, and about 8 inches in length. The filter pump is now turned on, and the stopcock of the shorter tube slowly opened. The barium carbonate remains on the asbestos, and the clear baryta solution which passes through is at once neutralised by the hydrochloric acid. When all the liquid has been filtered, the pump is allowed to act for a few moments so as to partially exhaust the bottle. The stopcock of the shorter tube is then closed.

Meanwhile 100 c.c. of distilled water, which always contains carbon dioxide in solution (compare Letts and Blake, loc. cit., 125; Walker and Cormack, this vol., 8, 11), is neutralised by adding to it phenolphthalein, a little barium chloride solution, and then baryta solution until an incipient pink colour is produced. Into this prepared washwater, contained in a small beaker, the end of the longer tube is