TEANSACTIONS OF SECTION B. 483 



obtained by tliem, nor made any difference between the values of the figures 

 deduced from operations on the large scale employed by Stas and those arrived at 

 on the small scale of other observers. Any sort of handicapping of methods is 

 no doubt a very difficult and delicate operation, and requires more than the judg- 

 ment of an Admiral Rous, but -without it the question whether the numbers 

 adopted are the best obtainable will always be an open one. It is, however, a 

 very noteworthy fact, that in almost every case the numbers deduced from fStas' 

 experiments, taken by themselves, coincide very closely indeed with the most pro- 

 bable numbers derived by the method of least squares from the whole of the re- 

 corded estimates. On the whole, Professor Clarke concludes that Front's hypothesis, 

 as modified by Dumas, is still an open question, that is to say, his final numbers 

 differ from whole multiples of a common imit by quantities which lie within the 

 limits of errors of observation and experunent. 



Let us tm-n again to the evidence aflbrded by our most powerfiU' instrument for 

 inspecting the inner constitution of matter, the spectroscope. A few years ago 

 Mr. Lockyer supposed that the coincidence of rays emitted by ditierent chemical 

 elements, particidarly when those rays were developed in the spark of a powerfid 

 induction coil and in the high temperatures of the sun and stars, gave evidence of 

 a common element in the composition of the metals which produced the coincident 

 rays. Such an argiunent could not be drawn from the coincidences imless they 

 were exact, and the identity of the lines could only be tested by means of spectro- 

 scopes of great resolving power. By the use of the well-known Rutherford gratings, 

 Yoimg in America had fomid that most of the solar lines which had been ascribed 

 to two metals were in reality double, and Dewar and I, worldng on the terrestrial 

 elements in the electric arc, had found the actual coincidences to be verj- few 

 indeed. These observations, even with Rutherford gratings, were delicate enough ; 

 but quite recently M. Fievez, of the Brussels Observatory, has brought to bear on 

 this question a spectroscope of unexampled power. By combining two of the 

 Astronomer Royal's highly dispersive half-prisms with a Rutherford gratmg 

 of 17,296 lines to the inch, he has obtained a dispersion quadruple — that of 

 Thollon's combination of prisms. Bringing this to bear on the smi he has mapped 

 the solar spectriun from a little below C to somewhere above F, on a scale one- 

 third greater than that of Vogel's map, and has not only confirmed the work of 

 Yomig, Dewar, and myself, but has resolved some lines which were not divisible by 

 such dispersive power as we had at command. This result cannot fail to shake our 

 belief, if we have any, in the existence of any common constituent of the chemical 

 elements ; but it does not touch the e-sidence which the spectroscope afibrds us, 

 that many of our elements, in the state in which we know them, must have a very 

 complex molecular structure. I cannot illustrate this point better than by the 

 spectra of two of our commonest elements, magnesium and iron. We have good 

 reason to think the molecule of magnesium to be as simple as that of any chemical 

 element, and we find its spectriun to be one of the simplest, consisting of a series of 

 triplets which repeat each other in a regular way and are probably harmonically 

 related, and of a comparatively small nmuber of single lines, of which also some 

 may be harmonics. The spectriun of iron, on the other hand, presents thousands of 

 lines distributed irregularly through the whole length, not only of the visible, but 

 of the idtra-violet region. Make what allowance you please for unknowii harmonic 

 relations, and for lines which are not reversible and may not be directly due to 

 vibrations of the molecules, we still have a number of vibrations so immense that 

 we can hardly conceive any single molecule to be capable of all of them, and are 

 almost driven to ascribe them to a mixture of differing molecules, though we have 

 as yet no independent evidence of this, and no satisfactory proof that any of this 

 mixture are of the same Mud as occur in other elements. 



M. Fievez's combination is a great advance in resolving power, but Professor 

 Rowland, of the John Hopkins University, promises us gratings not only exceeding 

 Rutherford's both in dimensions and accuracy of ruling, but ruled upon curved 

 surfaces so as to dispense with the use of telescopes and avoid all variations in 

 focussing the different orders of spectra. His instruments, if they come up to the 



Ii2 



