July 1 1, 1895] 



NATURE 



H7 



aluminium wire, and the deteriorating local currents in 

 the cell are avoided, if the glass tube exactly fits round 

 the aluminium wire ; but this desirable result has not yet 

 been perfectly attained, the liquid finding its way into 

 the tube after some considerable time. However, in 

 this way have been constructed cells which have re- 

 mained constant for about three weeks. 



In the figure, t: c is a cork in which the glass tube, n, 

 lontaining the aluminium wire at the end A and the 

 attached platinum wire, fits, this cork fitting lightly into 

 the side of the glass cell which contains the liquid. The 

 tube H passes close up to a quartz window, (j (^), cemented 

 to the cell opposite the cork C C. The light of the star is 

 received on the window, Q Q, and is made to fall on the 

 selenium layer at the end A of the tube li. .A. platinum 

 wire, p', is sealed into the bottom of the glass cell, and 

 conveys the charge taken by the liquid to one pole of the 

 electrometer, while the platinum wire l' conveys the charge 

 taken by the selenium to the other pole of the electrometer; 

 s IS a ground stopper at the top of the cell, where the 

 liquid is poured in. 



This cell is fitted into a holder which can be fi.\ed to a 

 telescope in place of the eyepiece ; and this cell-holder 

 allows of the adjustments which are necessary to bring the 

 point .\ to the position of the image of a star. 



This is the form of ])lu)toelectric cell with which, in 

 conjunction with Prof Fitzgerald and Mr. \V. E. Wilson, 

 1 measured the electromotive forces of the lights of 



Jupiter, Satum, \'ega, .Arcturus, Regulus, I'rocyoii, and 

 some other stars last .April, in .Mr. Wilson's observatory 

 at Uaramona, Westmeath. The telescope used was Mr. 

 Wilson's 2-feet reflector. 



In order to give a notion of the sensiti\eness of the 

 •cell to light, I may say that if an ordinary paraffin candle 

 is held at a distance of 9 feet fronr the window (.)y, it will 

 produce an electromotive force of about '03 volts ; or, to 

 put the matter differently, suppose an ordinarv quadrant 

 ■electrometer, of Clifton's pattern, charged so that a 

 ])aniell cell gives a deflection of 400 divisions on the ordin- 

 ary scale (placed at a metre distance,! ; then the light of 

 the candle at 9 feet falling on the photoelectric cell would 

 give a deflection of twelve di\ isions, and the deflection 

 varies inversely as the distance of the candle. 



Now the light of X'ega as concentrated in the 2-feet 

 telescope gives a slightly greater deflection than the 

 (of course unconcentrated) light of the candle: so that we 

 arc evidently dealing with easily measurable quantities. 



The cell is sensitive to all the' rays of the spectrum, but 

 the inaximum effect is produced by the yellow. It is 

 sensitive to rays considerably below the \ isible red and 

 beyond the blue. 

 _ The light of .Arcturus was found to give o'82 of the 

 K.M.F. produced by the candle at 9 feet; the light of 

 Saturn 056, which was also about the value of the light 

 of Regulus. Unfortunately neither .Sirius nor Capella, 



nor any star in Orion, nor any in the Great Bear, was 

 available for our observations ; but these we hope to 

 include, before long, in the list of measured stars. 



It will be observed that in this electrical measurement 

 of starlight we do not measure lurrents, but electromotive 

 forces — we do not use a galvanonietei, but an electro- 

 meter ; and an electrometer of small capacity was 

 specially constructed for these e.xperiments, with the aid 

 of the (iovernment grant dispensed by the Royal .Society. 



It is not desirable to allow the light to generate 

 currents: the electrical charges must be allowed to flow 

 back into the cell, so that it may not be temporarily 

 deteriorated during the observations. Hence the pre- 

 ference for the electrometer. 



The space at my disposal will not allow of my entering 

 into many details : but I may mention, in particular, the 

 importance of having the whole of the sensitive surface in 

 the cell covered by the light of the star. It matters not 

 to the value of the E.M.F. produced how far behind the 

 focal image of the star the sensitive surface, A, is 

 placed — provided that the image of the star just 

 covers the surface A. This is essential in all 

 photoelectric cells, and also in thermopiles ; and the 

 neglect of this condition may partly explain the 

 failure of attempts to obtain thermoelectric indications 

 from the stars and planets, although we should scarcely 

 expect success from methods which aim at measuring 

 merely a \ei-)- limited portion of the radiation I'viz. the 

 heat, or infra-red 1. The photoelectric cell integrates the 

 whole energy of the radiation on the sensitive surface; 

 and the sgtinre of the observed E..M.F. is the measure of 

 this incident energy, 



It is interesting to know how ihc p/iofelectric measures, 

 so far as they have gone, compare with the photometrU 

 measures of "magnitudes" hitherto employed by 

 astronomers. In the latter, if B and B' are the "bright- 

 nesses " of two stars of the magmitudes 111 and iii 

 respectively, we ha\ e by definition 



1 B 4 , ' 



III)', 



(I) 



This equation defines merely the diflerence of the 

 magnitudes, and the definition is quite arbitrary. 

 The essential things are B and B'. How are 

 they measured .' The photoelectric method says that 

 they are E^ and E'-, the squares of the electromotive 

 forces generated in a given cell by the lights of the two 

 stars. The ])hotometric method says that they are 

 measured by the thicknesses of certain interposed glass 

 prisms which extinguish the lights, or by polarising 

 apparatus which render the shades of the transmitted 

 lights "equal." Hence we may expect, perhaps, a fair 

 amount of agreement between the two methods, if we are 

 comparing two or more stars of the same colour. Thus, 

 in the photoelectric method, we have for any two stars 



III' - w = 5 iog,o -:-, (2) 



.A])i)lying this to .Arclurus and Regulus, and taking the 

 magnitude of the former as "2, we find the magnitude of 



Regulus to be f^ 



In .Miss Clerke's ".Svstem of the 



NO. 1 34 1, VOL. 52] 



.Stars ' (.Appendix), Regulus is quoted asi4, .Arcturus 

 being '2. 



Comparing in the same way I'rocyon and Regulus, the 

 latter being taken as of magMiitude i -33, the magnitude 

 of Procyon would be "46. Miss Gierke c|uotes Procyon 

 as of magnitude '5. 



But no agreement between the two methods is to be 

 expected when two stars of different colours are com- 

 pared. The photometric method of equalisation seems 

 to be just as meanmgless as the ordinary " grease-spot ■' 

 method of attempting to equalise a blue and a red light I 

 In this case the only intelligible comparison of two lights 

 consists in measuring the energies which they radiate per 



