maintains that the anilino-group is "negative," others may hold a contrary opinion; this question is immaterial to the present argument, since the general polar effect must be in the same direction, whatever the position of the substituent, and therefore cannot account for the fact that the a-anilinoacid is stronger, and the B-acid weaker, than propionic. The steric factor always increases the strength of an acid, since in the electrolytic equilibrium R.COOH R. COO'+H+ the unimolecular reaction from left to right is not affected by steric hindrance, whereas the bimolecular reaction from right to left is retarded thereby (solvation rendering the hydrogen ion large enough to be thus affected). Neither can the steric factor therefore account for the above inversion in B- compared with a-. This inversion would, as has just been shown, result from the quantitative factor by itself; and since tervalent nitrogen, as evidenced by its directing power in the benzene nucleus, has the greatest affinity demand, and the quantitative effect is therefore at its maximum, the resultant of all three factors also causes inversion. Professor Lowry suggests that the low constant of the B-acid might be explained by assigning to it the constitution of an inner salt. But this is excluded. An anilino-group is incapable of forming an inner salt with a carboxyl-group. In o-aminobenzoic acid the dissociation constant of the amino-group is only 15 per cent. less than in the corresponding methyl-ester; in m-aminobenzoic acid, though its configuration is unfavourable to inner salt formation, this difference is actually greater and in the same direction. In the anilino-aliphatic acids, on the other hand, the excess of strength of the a-acid over that of the B-acid is of an entirely different order, reaching for iso-butyric acid well over 1000 per cent. Moreover, inner salts invariably melt at very high temperatures, generally with decomposition, and are more soluble in water than in organic solvents. For instance, both a- and B-aminopropionic acids—which both form inner salts on account of their unsubstituted (by phenyl) and therefore very much more basic amino-groupdo not melt at all and are very easily soluble in cold water. But B-anilinopropionic acid melts at 59° to 60° (even lower than its a-isomeride (m.p. 162°) to which Professor Lowry does not assign an inner salt constitution), and is little soluble in cold water, but easily in organic solvents. All these facts serve equally to disprove assumptions, similar to those of Professor Lowry, made by Vorländer and by Adams (to which Professor Lowry has kindly called the writer's attention). While thus affording a striking confirmation of the above non-electronic theory, the anilino-acids--amongst othersfurnish a disastrous test for electronic theories. There we have, for instance : In other words, the acid which experience proves to be weaker than propionic acid should, in electronic parlance, possess carboxyl- oxygen with a particular avidity for electrons, and rice versa. And this, too, where we deal, for once, not with assumed, unproved electrons, but with a real, demonstrable electric charge. In the other series of constants discussed by Professor Lowry, most of the values differ relatively little, and in no case is there a substituent with such an outstanding quantitative effect as that of tervalent nitrogen. In such cases the application of all three factors cannot leave the quantitative effect in such a predominant position, as a glance at the author's papers of 1909 and 1910 will show, where all three factors have been applied to numerous constants (including the anilino-acids), by a concordant method throughout which automatically excludes arbitrary adjustments. Professor Lowry proves that the application of the quantitative factor alone does not yield general agreement—a fact not only in conformity with the author's theory, but postulated by it. *Lieb. Annalen, cccxx. p. 99 (1902). Journ. Amer. Chem. Soc. xxxviii. p. 1508 (1916) It can be shown that when all three factors are applied, all the constants examined by Professor Lowry are in agreement with the theory, subject to the following remarks: 1. For vinylacetic acid, Fichter's value (3.8 × 10−5) should be replaced by Zelinsky's re-determination (4·65 × 10−5). 2. As to unsaturated acids, in general, one must agree with Professor Lowry that there is uncertainty where only one of the two cis-trans-isomerides is known. But it can be shown that this affects only aß-unsaturated acids. 3. That according to the author's theory n-valeric acid must not be stronger than n-butyric acid, is also a point on which Professor Lowry is quite correct. All the published comparative determinations for these two acids have therefore been examined. There are three from Ostwald's laboratory: The discrepancies are therefore considerable, but only in No. 2 (Francke) have both acids been carefully purified. There is, at any rate, no reason to conclude that n-valeric is stronger than n-butyric acid. But butyric, in spite of the positive effect of methyl, should be and is somewhat stronger than propionic acid (steric effect of B-substituent, see 1909 paper, p. 729). To summarize: organic facts can be concordantly coordinated on on the basis of varying affinity demand in conjunction with general polarity and steric hindrance. Electronic bonds, however, are confronted with serious difficulties by the facts of organic chemistry as well as by the postulates of physics. Fleet, Hampshire. L. A New Form of Electrometer. [Plate II.] S was shown in a paper published some years ago† numerous problems in astrophysics could be solved if a convenient photoelectric method were developed for measuring the amount of light received from stars, nebulæ, comets, etc. For this purpose the photoelectric cell is best fixed just behind the focus of an equatorial telescope and the current produced by the light which falls upon the cell is measured by means of an electrometer. In order to keep the electrostatic capacity of the system as small and constant as possible, it is desirable to fix the electrometer and the cell. rigidly together. This also facilitates drying and thus diminishes any leakage. This arrangement involves considerable difficulties, however, since no existing form of electrometer maintains a constant sensitivity, and still less a steady zero, when moved about at the end of an equatorial telescope. The object of this paper is to describe a new form of electrometer which is free from these defects. Since it is capable of a sensitivity as great as that of any existing electrometer, although its period is less than one second and its capacity less than 2 cm., it may be useful for many laboratory purposes even where independence of the zero and sensitivity to tilt is not essential. The principle of this electrometer is the same as that of the quadrant electrometer; it consists essentially of a needle, which in this case really has the shape of a needle, suspended at its centre on a torsion fibre in such a way that it can rotate between four cross-connected plates which take the place of the ordinary quadrants. The important innovation is that the torsion fibre is fixed at both ends under tension so that the centre of rotation of the needle is fixed, and its rotation can therefore be determined by observing the motion of one end through a microscope. This renders a mirror unnecessary, and, consequently, the moment of inertia of the moving parts can be reduced to a minimum. Hence a very small torsional restoring force may be used without increasing the period unduly. The most suitable dimensions can be found by considering *Communicated by the Authors. Monthly Notices R. A. S. lxxix. 5. p. 343 (1919). Phil. Mag. S. 6. Vol. 47. No. 279. March 1924. 2 P 4 the sensitivity in terms of the dimensions of the various parts. If r is the radius of the torsion fibre, l1 its semilength, and n1 the torsion modulus, the restoring couple for unit angular displacement is F。= If r is the effective radius of the needle, l, its semi-length, and p2 its density, the moment of inertia I is prl. Hence the free period in vacuo will be πηγη If an earthed conductor of capacity C in an electric field F be moved a distance & in the direction of the field a charge C.8.F will be induced upon it, assuming its reaction on the field to be negligible. Hence if c be the capacity of the needle per unit length, rotation through an angle will induce a charge leFdl in each element dl. The electric field therefore tends to make the needle unstable, the couple to which it gives rise being 2FcFdl, which reduces for 0 unit angular displacement to 3,F2,3. Hence the effective restoring couple I will be To diminished by this electric couple, i. e. T= t πηγ 4 - col3F2. The period in the field therefore, if damping be neglected, will be If a charge at potential v is communicated to the needle, each element di will contain a quantity of electricity vedl. The turning moment in a field F will therefore be taking account of both arms of the needle. Hence the will be displaced by a charge at unit potential through a distance |