and indirectly in the intervening space, the body of the arc which they would have to cross, and failed to find them there. I feel, therefore, that the failure of this test is a serious objection to his theory.

With regard to my suggested explanation of the repulsion of the poles I am afraid that I cannot have made my argument quite clear. Except close to the electrodes an ion travels with an average velocity determined by the field and the frictional drag of the ion on the gas. We are agreed that the resultant effect on the electrodes of this process is nil. But in the initial stage of its motion an ion starting from rest at the electrode surface must accelerate in order to acquire this average speed--in other words it is acquiring more momentum than it is imparting to the gas. I still maintain therefore that the electrode must feel the reaction whether the ion reaches its final average velocity rapidly or slowly. And whether an assumption such as I suggested, namely that the negative ions are charged atoms (which leads to a quantitative agreement with experiment) is justified or not, a reaction of some magnitude must be there.

Professor Duffield appears in one case to dislike my terminology. I do not quite see his reason for doing so. When ions moving from an electrode through a gas at atmospheric pressure exert a drag upon the gas they tend to set up behind them a region of lower pressure into which other gas or the electrode itself, if free, will move. I feel myself that the term 'suction' is a convenient and brief term for expressing this process.

University of Bristol.

I am,
Yours faithfully,

A. M. TYNDALL.

June 11th, 1923.

XXXV. The Octet Theory of Induced Alternate Polarities: the Domains occupied by Octet-stable and Octet-unstable Centres. By RONALD FRASER and J. E. HUMPHRIES, B.A., M.Sc., Chemistry Department, University of Aberdeen*.

THE authors have recently (Chemical News, 1923, cxxvi. pp. 241, 257) applied the Lewis-Langmuir theory of covalence to the elucidation of the principles underlying numerous groups of facts and empirical rules in organic chemistry. The discussion rests primarily on an electronic

* Communicated by Prof. A. Findlay, M..A., D.Sc.

interpretation of Lapworth's Principle of Induced Alternate Polarities, similar to that advanced by Kermack and Robinson (Jour. Chem. Soc. 1922, cxxi. p. 427). We differ from these authors, however, in postulating an electron transference as between stable and unstable octets as a necessary accompaniment of the manifestation of induced alternate polarity; that this is not their view was made clear by Professor Robinson at the Hull Meeting of the British Association, 1922. There appears to us to exist a certain amount of justification for our assumption: it is now widely recognized that there is no essential distinction between polar and non-polar compounds (G. N. Lewis, Jour. Am. Chem. Soc., 1916, xxxviii. p. 762; Briggs, Jour. Chem. Soc., 1921, cxix. p. 1879; Kermack and Robinson, loc. cit.; Stieglitz, Jour. Am. Chem. Soc., 1922, xliv. p. 1293); in an extreme case of the attainment of octet stability, such as that of the chlorine ion in the strongly polar compound NaCl, there is no doubt that electron transference has taken place; and it seems only in accordance with continuity to suppose that a chlorine atom in, say, monochlorbenzene has drawn an electron to itself from the carbon atom to which it is attached.

Further, in the extreme case of octet stability where the octet can exist free as an ion, its reactivity is exceedingly great; and we have found, in an examination of numbers of organic compounds in the light of the octet theory, that reactivity and octet stability run parallel. Perhaps it would be carrying the argument too far to assume ionization, even momentary, in such cases; but in any event it would appear that the more nearly a substituent approaches octet stability, the more it should spread from the nucleus. In view of the impossibility at present of determining the relative sizes of stable and unstable octets, we have been led by the above reasoning to seek a vindication of our opinion by considering the domains occupied by stable and unstable centres attached to the benzene nucleus.

Possible methods of attack include: (a) Viscosity measurements. An application to the present problem is immediately suggested of the methods employed by A. O. Rankine to determine the sizes of the protuberances formed by hydrogen atoms on molecules containing hydrogen (Rankine, Proc. Far. Soc., 1922, xvii. Part iii.). Unfortunately, the benzene derivatives we should wish to examine are liquids or crystalline solids, and an examination of these in the gaseous state, as would be demanded by Rankine's methods, presents considerable experimental difficulties: (b) Critical data. It has been pointed out by Chapman (Phil. Trans. Roy. Soc., 1916,

ccxvi. p. 279), that molecular sizes deduced from viscosity measurements and from the be of van der Waals' equation are in excellent agreement, for molecules containing up to three atoms. The bodies we should wish to examine contain six or more atoms, so that in such case too much faith could not be placed in the agreement noted by Chapman. There seems little doubt, however, that the foregoing are the best methods available; as has, indeed, been remarked by Wöhlisch (Z. Elektrochem., 1921, xxvii. p. 295).

Owing to the time which must necessarily elapse before the results of extended inquiry along these lines may be available, we venture to think that the interest of the point at issue warrants our presenting here certain data which appear to lend support to our view, namely, that electron transference as between stable and unstable octets accompanies the manifestation of induced alternate polarity. We have made a re-examination of the work of Jungfleisch (Compt. Rend. 1867, lxiv. p. 911), and Le Bas (Phil. Mag. 1914, xxvii. p. 988), on the molecular volumes of various chlorbenzenes; we have taken merely the average volume of the chlorine atoms in each several compound, as we feel that the additive principle of Kopp can be pushed too far in investigations of this kind. Below are placed the ideographic representations of the compounds, with the state of octet stability and instability of the chlorine atoms indicated by - and respectively. The figure beneath each compound is the average volume of the chlorine atoms, as computed

It will be seen

from the data of the authors mentioned. that, in every case, the introduction of an octet-stable (-) chlorine causes an increase in the average volume of the chlorine atom, that of an octet unstable (+) chlorine a decrease, just as one would be led to expect on our view.

There is a further result of considerable interest; it might, in fact, be regarded as something in the nature of a crucial test. It will be observed that in all the compounds examined by Jungfleisch, certain of the chlorine octets are

stable, others unstable. Now, in the case of m-dichlorbenzene, we regard both chlorine atoms as in a state of octet stability; they are, therefore, reinforcing each other's tendency to spread from the nucleus; and hence, if our reasoning is correct, the average volume of the chlorine atoms in this compound should be greater than in any of the former cases where no such reinforcing effects existed. We have calculated the molecular volume of m-dichlorbenzene at the boiling-point (172°C.) from its density at 0°C. (1·307; Beilstein, vol. ii. p. 44) by means of the formula due to Le Bas (Proc. Chem. Soc., 1914, xxx. p. 86):

273

do/d=1+c(1-7), where c=0·46.

The following figure gives the state of polarity of the constituent atoms of m-dichlorbenzene, and the average volume of the chlorine atoms, which may be taken to be correct to within 1 per cent. (Le Bas, loc. cit.). It should be compared with the first group.

Fig. 2.
í

21.5

The result is in accordance with prediction, the average volume of the chlorine atoms being about 3 per cent. greater than the largest value (20·9) obtaining in the polysubstituted members of the former group. (Monochlorbenzene appears to stand apart from the general run of the results.)*

We consider the above data to give at least a pointer towards the correctness of our ideas, and, as such, to be of some interest; but it would be unwise to make any general statement at the present stage, until more extended results may be available for discussion.

We are glad of this opportunity of expressing our gratitude to Professor Alex. Findlay for his interest and constant encouragement.

Fry (The Electronic Conception of Valence and the Constitution of Benzene,' 1921, p. 177 seq.) has advanced a different interpretation of the results of Jungfleisch; but his theory leads to many inconsistencies, and it does not seem possible to predict, on the basis of his conceptions, the decided increase in the average volume of the chlorine atoms in m-dichlorbenzene.

THE

XXXVI. A New Phenomenon *. By MARSHALL HOLMES. HE phenomenon to be described in this paper is thought to be new, and although it has a direct bearing on a larger question now being investigated, it is perhaps sufficiently interesting to warrant publication by itself.

The apparatus used was arranged in the manner shown:

B

C

A is a polarimeter; B is the polarimeter cell containing a substance suitable for the investigation; C is an electromagnet with the bobbins arranged across the bed; D is the source of light; and E is the switch. The nicols are arranged so that the polarimeter field appears uniform at first.

To obtain the phenomenon, the polarimeter cell is filled with a suitable liquid-glycerine was generally chosenand placed in position in the polarimeter. The polarimeter field still appears uniform as glycerine is optically inactive. The current through the magnet is now switched on and the magnetic rotation produced causes part of the field to appear dark. If the adjustment of the polarimeter is left unaltered, and the current is now switched off again, attention being meanwhile steadily directed on the image, it will be seen that it first fades rapidly away and then just reappears with the light and dark parts reversed before the field finally becomes quite uniform, so that if the magnetic rotation has originally been dextro the reversed rotation is lavo, or if it has been levo the change is to deatro. What happens with most substances is that the field becomes quite uniform immediately the current is switched off, and it remains uniform. With glycerine, however, such is not the case.

The phenomenon takes place and disappears very quickly, *Communicated by the Author,