values of N connected with Mohler and Foote's voltages. lie either on, or very close to, this line. The line Kas was observed by Hjalmar in the case of aluminium, and by computing L from the relation given above and ascertaining the voltage connected by the quantum relation with the wave number so determined, the value 49.1 volts is obtained. This value is rather higher than the value (42 volts) which we determined experimentally for aluminium, so that it seems unlikely that our experimental value is to be associated with this hypothetical L limit. Wentzel has shown that the additional K emission lines observed by Hjalmar on the short wave-length side of Ka are emitted by atoms which have lost more than one electron, and which have therefore absorbed more energy than that corresponding to the ordinary K absorption limit. If this is the case, then by taking the difference K-Kas we do not obtain a quantity which has any true physical significance in the sense of an absorption limit. An L absorption limit on the short wave-length side of the usual limits might be expected if the simultaneous removal of two electrons from the L group took place, but it is difficult to see how an absorption limit on the long wavelength side of LII, III can be accounted for by such multiple ionization. The value (42 volts) obtained in our experiments must be connected with a genuine absorption limit, and not simply with a quantity which is calculated from the difference between a K absorption limit and a K emission line which corresponds to some exceptional state of the atom. For the elements titanium (=22), vanadium (2=23), and chromium (≈=24) Fricket found photographically an absorption discontinuity lying on the long wave-length side of the principal K discontinuity, and it has been suggested by Coster that this anomalous discontinuity is to be connected with the removal of an electron from the K group to one of the incomplete M sub-groups. Such a suggestion receives support, as Coster has pointed out, from the discovery by Hjalmar of an emission line which he calls KB', which has exactly the same wave-length as the anomalous discontinuity. The authors therefore think it probable that the critical stage which they obtained at 42 volts, in aluminium, and the low values found by Mohler and Foote for the elements sodium, magnesium, phosphorus, sulphur, and chlorine correspond to the displacement of an electron within the atom from the second L sub-group to one of the *G. Wentzel, Ann. der Phys. lxvi. p. 437 (1921). ↑ H. Fricke, Phys. Rev. xvi. p. 202 (1920). 740 Excitation of Characteristic X-rays from certain Metals. incomplete M sub-groups. According to Bohr's theory, the first M sub-group in aluminium contains two electrons and the second M sub-group contains one electron. The question arises as to which of these two incomplete M sub-groups receives the displaced electron at the 42-volt stage. If the electron enters the second M sub-group, we should expect LII, III-L(anomalous) to be comparable with the ionizing potential of the element of next higher atomic number. The difference between our values for LII. III and Lanomalous) is 66 volts-42 volts=24 volts, however, a value which is considerably higher than the ionizing potential of most elements, so that it seems unlikely that the critical value at 42 volts for aluminium can be accounted for in this way. Moreover, a transition from the second L sub-group to the second M subgroup would not be in accordance with the selection principle, since the azimuthal quantum numbers of these two sub-groups are the same. It would appear then, that 42 volts corresponds to the energy necessary for the removal of an electron in the aluminium atom from the second L sub-group to the first M sub-group". Summary. Investigations of the voltages connected with some of the longer wave-length absorption limits have been made for the elements aluminium, iron, nickel, copper, and zine by the excitation potential method. The following critical values have been obtained : For aluminium... 42 volts, 66 volts, and 107 volts. By extrapolating the Moseley curves for the various absorption limits to low atomic numbers, the lower of the two * It has been pointed out to the authors by Prof. Bohr and Dr. Coster that for elements of atomic number below about 24 the values of √ for LII, III given in their tables cannot be relied upon to the same extent as the values for higher atomic numbers, on account of the various electrical conditions of the atoms in the different compounds used in the experiments from which the data for the calculation were obtained. The possibility that the critical value of 42 volts for aluminium should be associated with the LII, III level, and the critical value 66 volts with the Lr level, while the highest value should be associated with some doubly ionized condition of the aluminium atom, cannot, therefore be entirely ruled out on the evidence at present in existence. values obtained for each of the metals, iron (z=26), nickel (28), copper (29), and zinc (30), was found to be associated with the MII, III level, and the higher of the two values obtained for each of these elements was found to be associated with the Mr level. The changes in the slope of the MI and MI, III curves in passing through the observed points support the view that a change in the constitution of the M electronic group is in progress as we pass from one of these elements to another. The results suggest that the development of the second M sub-group from four electrons. to six electrons, as required by Bohr's theory, does not commence before the element cobalt (27), but that the similar development of the first M sub-group commences for a lower value of the atomic number. By extrapolating the Moseley curves for the L absorption limits in a similar way, the two higher values found for aluminium have been connected with the L and LII, III levels respectively. From analogy with the anomalous K absorption limits found by Fricke for titanium, vanadium, and chromium, which Coster suggests are due to the displacement of an electron from the K group to an incomplete M sub-group, it seems probable that the lower value (42 volts) found for aluminium is to be associated with the displacement of an electron from the second L sub-group to the first M sub-group. LXXXIV. Vacuum Grating Spectrograph and the Zinc Spectrum. By R. W. WooD, Professor of Experimental Physics, Johns Hopkins University*. [Plate XI.] Tonale quatings mounted in vacuum spectrographs HE reproductions of spectrograms obtained with which have been published during the past five years appear to indicate that higher resolving power in the short wave-length region is very much to be desired, if accurate determinations of wave-length are to be made. During the past winter I have made some preliminary investigations with two instruments of this type which have been constructed in the shop of the University. In view of the present grating situation it appears to be worth while to publish a brief statement of what may be * Communicated by the Author. expected of short-focus concave gratings, which can now be produced on the Rowland engines as desired. A good deal of time was spent in getting the spectrograph into proper condition, as much trouble was found owing to the presence of absorbing vapours, and at the close of the University year but two satisfactory spectrograms had been obtained, one of carbon and one of zinc, with carbon as an impurity. The lines on these two plates, however, were sharper than any that I have ever seen, and the focus appeared to be perfect over the entire range of the plate. A three-fold enlargement of the zinc spectrogram is reproduced in Plate XI. a in coincidence with a corresponding portion b, of the spectrum enlarged to the same scale from the reproduction accompanying Sawyer's paper on the Zinc spectrum, published in the Astrophysical Journal' for December 1920. Certain groups of lines, enlarged 7.5 times, are reproduced. at e, and immediately above these, small regions of these groups, enlarged 34 times, are given. It is only in these last that the lines show a width comparable with the width of the lines in Sawyer's spectrum. The corresponding portions of spectrum, in the series of enlargements, are indicated by brackets. This spectrum will be discussed more fully presently. The Spectrograph. The spectrograph was constructed along lines similar toʻ those indicated in Prof. McLennan's paper, published recently in the Proceedings of the Royal Society. Certain modifications, intended to facilitate the operation and adjustments of the instrument, were introduced, but will be passed over without comment, as, in the opinion of the writer, the best type of vacuum spectrograph is yet to be designed. Through a misunderstanding on the part of the mechanic, the large conical ground joints were put together with alcoholic shellac, the vapour from which was doubtless responsible for the failure of the instrument to record anything below wave-length 1600 for several weeks. With continued operation, however, the range gradually increased until a group of lines at 834 appeared, the limit obtained up to the present time. With more rapid methods of exhaustion, however, and continued operation, matters will undoubtedly improve. Up to the present time I have given but little attention to the design of the instrument, being occupied chiefly with the question of the best type of grating to employ. Thus far but one grating has been ruled, but as it appears to yield spectra of high quality with comparatively short exposures, a brief description of it may not be out of place. The Concave Grating. Owing to the difficulties found in securing a sufficiently constant temperature in the dividing-engine room, practically no gratings have been ruled on the Rowland machines. since the removal of the Physical Laboratory to its temporary quarters in the Electrical Engineering Building six years ago. The walls of the small room are exposed to the outer air on two sides, and the gas-heated radiator controlled by a thermostat was found to be useless, as the flame was frequently blown out by air blasts from outside. Having need of a short-focus grating for a recently constructed vacuum spectrograph, I took up the problem of securing proper conditions for ruling last autumn. The gas radiator was removed, and the inlet and exhaust pipes through the wall closed. In its place was installed a small gas stove of sheet iron, heated by a ring of small bat-wing burners. This type of flame was chosen in preference to the bunsen flame, as it can be turned down indefinitely without snapping back. The stove was provided with a chimney of sheet iron about 12 feet long, which passed out through the wall into the corridor. The gas supply was regulated by a toluene thermostat, the rising mercury column closing a glass tube perforated with a very minute hole a little above its open end. This hole served as a bypass for gas in sufficient quantity to maintain flames about 2 cms. in height, furnishing enough heat to keep the room several degrees above that of the outside air in the warmest weather. This arrangement has given perfect satisfaction, the temperature in the room having held constant to within 0°.2 throughout the winter and spring. Except when sudden and extreme temperature changes occurred outside, the changes in the room were not over 0°∙1. While this is satisfactory for short-focus gratings of moderate resolving power, a better control was desirable for ruling very large gratings of high resolving power. To meet this requirement a large grid of very fine insulated resistance wire was set up in the glass cage in which the |