Feb. 13, 1879] 



NATURE 



Z'^Z 



length. From the fact that this error runs through the entire , 

 paper, it would almost seem as if the author was not aware of 

 the distinction between wave-lengths and scale numbers. Using 

 Angstrom's scale he confuses wave-lengths with ten millionths 

 of a millimeter ; whereas in the case of D for example, the 

 wavelength is nearly 6,000 times greater. If the author really 

 means what he says, he asserts that the wave-length of the mean 

 x-ay of the spectrum is one two-hundred and fifty-millionth of an 

 inch instead of about one forty-thousandth, as we know it is. 



Second, the author deems it of the greatest importance in the 



iireparation of his solar photographs to use reflected rays exclu- 



ively; saying, p. 256, last line, *' at no time did the solar rays 



^ass through glass ; all error that might arise during refraction 



as thus avoided." After this virtual condenmation of the use of 

 - efraction at all, he not only uses for comparison Angstrom's 

 wave -lengths made with achromatic lenses and a refracting grat- 

 ing, constructing even his chart upon them as a basis, p. 258, line 

 7, " the values assigned to the wave-lengths in this chart are 

 those of Angstrom " ; but the very spectrum of oxygen by which 

 the coincidence of the lines of fliis element with those of the 

 sun spectrum were to be established, was photographed with 

 glass prisms and achromatic lenses. 



Thhxl, the author states that the prisms with which the spec- 

 trum of oxygen was photographed were adjusted "to the 

 minimum de^^ation of D'." Supposing D^ to be meant, this 

 precaution, which gives the appearance of extraordinary accuracy 

 to the adjustment, is practically an impossibility with the appara- 

 tus employed. Minimum deviation of the D line as a whole could 

 not imder these circumstances be distinguished from that of either 

 of its components, nor could that of D^ be distinguished from 

 that of Dj. Moreover, it is difficult to understand why he 

 adjusts to minimum deviation for LK and not for G, near which 

 the work is to be done. Instead of D', the line for which his 

 apparatus was adjusted should have been chosen in the photo- 

 graphic portion of the spectrum, for example between G and H. 



Fourth, on page 265, line 25, the author says that this "is 

 a problem not to be solved by the comparison of two spectra 

 of small dispersion." Hence it is a matter of some surprise 

 to find that in getting his oxygen spectrum, he uses only "two 

 flint glass prisms of 60," and for objectives, " achromatics of 

 ten inches focus. " The bright line spectrum of oxygen taken 

 by Henry Draper, which the author in this paper inferentially 

 attacks, was made, as we find on examination, with a direct 

 vision battery of nine prisms and an observing telescope of forty- 

 two inches focal length. The original negatives taken with the 

 latter apparatus must consequently have been eight or nine times 

 as long as the author's ; and even these were none too large for 

 the proper solution of the question. 



Fifth, the author seems to have attempted to compare together 

 a diffiraction spectrum of the sim with a prismatic spectrum of 

 oxj'gen. Such a comparison, by the method adopted, is mani- 

 festly of no value. Owing to the irrationality of dispersion of 

 various refractive media it is an extremely difficult thing to com- 

 pare accurately two prismatic spectra of different kinds. But 

 the matter rises to an absurdity when a comparison is attempted 

 between a grating spectrum and a prism spectrum. The graphic 

 method, employed to supplement the direct method, does not 

 appeal" to help the comparison, since the author nowhere gives 

 both co-ordinates to the curve constructed. 



Sixth, it is more than questionable whether the measurements 

 of the solar lines actually made by the author are capable of the 

 accuracy he assigns to them. The values in his table of wave- 

 lengths are given to one hundredth of a division of Angstrom's 

 scale. As the author says on p. 257 that each division of this 

 scale, which is one millimeter, was enlarged to five millimeters 

 upon the paper scale on which the photographs were projected, 

 to measure to one hundredth of a scale-division would require the 

 measurements on the screen to be made to one-twentieth of a 

 millimeter or the one five-hundredth of an inch, about ; a degree 

 of refinement highly improbable under these conditions. More- 

 over the accuracy of the results of such measurements is seriously 

 impaired by the variation in the position of the lines on the 

 screen, due to the fact that the large number of negatives (eight 

 or nine apparently), required to give the whole photographic 

 spectrum, must, unless special precaution was taken (of which 

 there is no evidence), have been made with glass of different 

 thicknesises. When projected in the lantern, this variation in 

 thickness would necessitate a change in focus and so cause a 

 change in the minifying power. TTie smaler sizes of photo- 

 graphic glass vary in thickness from one to two millimetars. 



Consequently the displacement of the lines due to the difference 

 of magnifying power arising from this cause would exceed con- 

 siderably the limit of measiurement, which, as above stated, was 

 the one five-hundredth of an inch. But another and a more 

 serious cause of inaccuracy must here be pointed out. From the 

 data given by the author, it may readily be calculated that his 

 original photographs of the oxygen spectrum, taken with two 

 prisms of 60° and with lenses of ten inches focus, could not have 

 been over half an inch long in the region firom G to H. Since 

 Angstrom's chart from G to H is sixteen inches long, the author's 

 spectrum would have to be magnified thirty-two times to make it 

 the size of this. But as each millimeter of Angstrom's scale 

 was made five millimeters on the author's scale of measurement, 

 the original negative as thrown on the screen must have been 

 magnified one hundred and sixty diameters. Any one who has 

 worked at all in spectrum photography, knows thikt it is utterly 

 futile for piurposes of measm-ement to magnify a photograph 

 taken under these circumstances, as much as this, since then tie 

 size of the sUver grains becomes larger than the details of the 

 pictiure. In the absence of any precise statement the reader has 

 to make the calculation for himself ; but the figures above given 

 caimot be far astray. 



Seventh, there is only an appearance of accuracy when the 

 attempt is made to fix the position of the oxygen spectrum lines 

 to hundredths of one of Angstrom's scale divisions. The pro- 

 jection method by which his solar lines were measured, has 

 already been proved inadequate. And as to the method of 

 graphical interpolation, used as auxiliary to the lantern, it does 

 not appear that, as used by the author, it was capable of any 

 such accuracy as that claimed. In constructing the curve, the 

 iron lines are taken with Angstrom's values for the wave- 

 lengths ; but these, though estimated to tenths, were read only 

 to whole divisions of the scale. Moreover, only forty-seven 

 iron lines were used in all, or one to every eleven scale divisions ; 

 the reading being to one one-hundredth of a scale division, or 

 1, 100 numbers to one iron line. Since the author measured no 

 wave-lengths directly, he was obliged to construct a considerable 

 "portion of the curve from the wave-lengths of oxygen and air 

 lines already given by various authorities." These values were 

 taken, p. 258, from Watt's " Index of Spectra." On referring 

 to this book, the values are given only to the units place. And 

 even then, discrepancies amounting to from three to five entire 

 units, or from three hundred to five hundred times the author's 

 limit, appear in the wave-length as given by the various authors 

 relied" on for the measurements employed in the paper before us. 



Eighth, the author nowhere states the peculiar character of 

 the lines in the oxygen spectrum and appears not to know that 

 they have any. He has apparently taken it for granted that the 

 lines of oxygen are intrinsically as sharp as the lines of the solar 

 spectrum. But this, at least in many instances, is known not to 

 be the case. Consequently it is quite impossible to measure the 

 oxj^en lines as accurately as the solar lines, and even these, as 

 has been shown, ^cannot be measured to the accuracy which the 

 author claims. Angstrom himself admits that there may be an 

 error of one-tenth of a division in his scale numbers. 



It would seem sufficiently obvious from what has been said 

 that the results given in this paper are entirely vitiated by the 

 errors of method and of experiment which it contains. The 

 author must not be confounded, because of the similarity of 

 initials, with the distinguished investigator, Dr. J. W. Draper. 



UNIVERSITY AND EDUCATIONAL 

 INTELLIGENCE 



University education for women may be taken as firmly 

 established in England. The names of nine out of eleven female 

 candidates have just been posted at Burlington Gardens as 

 having passed the winter matriculation examination of the 

 University of London, half-a-dozen in the honours division, 

 besides three more in that next below, and are thus now 

 on the high road to its B.A. degree, on the occasion of its 

 first decorating their sex three academical years hence. From 

 two interesting articles in the Daily News it is seen that 

 Girton and Newnham Colleges have attained to unexpected 

 success. The former is so much too small for the number of 

 students that yearly flock to it that it is to be nearly doubled m 

 size. At both institutions the students work much more 

 earnestly than the average student of the hitherto privileged 

 gcx, and the examinations passed, at least at Girton, would 



