May 2. iSS^I 



NATURE 



v^ 



INWALDS MATHEMATICAL SPECTRUM 

 ANALYSIS. 



'HE following interesting criticism of Dr. Griinwald's recent 

 work on the mathematical spectrum analysis of various of 

 the elements, by Joseph S. Ames, of the Johns Hopkins Uni- 

 versity, appears in the February number of the American 

 Chemical journal : — 



"Dr. Anton Grimwald, Professor of Mathematics in the 

 Technical High School at Prague, has given his theory of spec- 

 trum analysis in the following papers :— (i) ' Ueber das Wasser- 

 spectrum, das Hydrogen-, und Oxygenspectrum,' Astronomische 

 Nachrichten, No. 2797, 1887, and Phil. Mag., xxiv. 354, 

 1887 ; (2) ' Math. Spertralanalyse des Magnesiums und der 

 Kohie,' Monatshefte fiir Chemie, viii. 650, IVietier Sitz. 

 Perichte, 2 Abth. xcvi., 1887; Phil. Mag., xxv. 343, 1888 

 (abstract); and (3) 'Math. Spectralanalyse des Kadmiums,' 

 Monatshefte fiir Chemie, ix. 956. 



"His aim is to discover relations between the elements by 

 tracing connections between their spectra, and thus to arrive at 

 simpler, if not fundamental, 'elements.' He considers the lines 

 in the spectra of two substances, say A and B. If he finds a 

 group of lines in the spectrum of A, which, on multiplication 

 with a simple numerical factor, give line for line a group in the 

 spectnim of B, he assumes that A and B have a common com- 

 ponent. This factor, which transforms the one group into the 

 other, is, he says, the ratio of the volumes occupied by the 

 common constituent in unit volume of the two substances. 

 Thus, let c be common to A and B, and" let it occupy the volume 

 \a\ in unit volume of A, and \b\ in unit volume of B ; then the 

 factor which transforms that part of the spectrum of A due to 

 f into that of B, also due to c, is \bii{a\ It is not difficult 

 to find relations between the spectra of different substances ; and, 

 accepting Dr. Grimwald's hypothesis as to the transforming 

 factor, we can deduce formulas for the elements. For exam[)le, 

 in ihe hydrogen spectrum there are two groups of lines, [a] and 

 [/'], which, when multiplied respectively by \% and \, give cor- 

 responding groups in the spectrum of water, and, since in water 

 hydrogen occupies § of the volume, we have the equations 



W + \l>\ = I 



which gives hydrogen the composition ba^. For reasons which 

 depend' upon solar physics. Griinwald calls the substance a 

 coronium, and /' helium. Further, he says that all the lines in 

 Hasselberg's secondary spectrum of hydrogen can be changed 

 into water-line by multiplving by \ ; which shows, according to 

 his theory, that the modified molecule H^ occupies in H2O half 

 the volume it does in the free condition. He finds that oxygen 

 has the composition Wb^b^Cr)^. where r is a new substance. In 

 his last paper, however, Dr. Griinwald states that he has proved 

 <■ to be nothing but a in a different state of compression. 



" He adopts the spectmm of water, i.e. of the o.xyhydrogen 

 flame, as a standard, and is then able to give various criteria by 

 means of which the primary elements;? and b may be recognized. 

 Among them are the following: If A is the wavelength of any 

 line produced by a as it exists in hydrogen, 1%K, \\, ^5 A. will each 

 be the wave-lengths of any line of the water-spectrum, and if A is 

 the wave-length of any line produced by b as it exists in hydrogen, 

 ^A will be the wave-length of a line of the water-spectrum. 

 Applying his criteria to magnesium, carbon, and cadmium, he 

 finds that they are made up entirely of « and b in various states 

 of compression. For instance, one group of li .es in the cadmium 

 spectrum is transformed into a group of b by the factor |, another 

 group is identical with a group of b, and so on. But the group 

 of lines of shortest wave length is transformed into a group of 

 b by the factor \ ; and cidmium falls in the seventh row of 

 Mendelejeffs table. Similarly, the group of lines of shirtest 

 wave-length of zinc is transformed into a group of /^ by the factor 

 \, and zinc is in the fifth row of the table. Dr. Griinwald findi 

 in this a general law which he verifies in the cases of Al, Si, Fe, 

 Cu, Zn, As, Sr, Ag, Cd, In, Sn, Sb, Te, Ba, Au, Hg. Tl, Pb, 

 and Bi. He farther connects the lines of greater wave-length 

 with the substance a ; and, as in all cases so far tried all th; lines 

 can be deduced from the-ie two substances, he is leJ to believe 

 that all the so called elenints are com'oun.ls of the priniry 

 elements a and b. 



"It is unfortunate that D •. Griinwald has not published a 

 complete list of the lines characteristic of a ani b, for until this 



is done his theory cannot be accurately tested. There are two 

 distinct questions to be answered : (i) Are there any numerical 

 relations connecting the spectra of the elements? and if so, (2) 

 what is the meaning of the fact? Cornu, Deslandres, and 

 others have long since answered the first question for us, but 

 whether Dr. Griinwald's answer to the second is correct or not 

 depends upon the completeness with which the numerical rela- 

 tions hold for the entire spectra of the substances. It is here 

 that Dr. Griinwald's work can be criticized. 



"As noted above, the spectrum of the oxyhydrogen flame is 

 used to test the existence of lines belonging to a and /'. By far 

 the most accurate and complete determination of this spectrum 

 is that of Liveing and Dewar (Phil. Trans. 1888) ; but this does 

 not always answer Dr. Griinwald's purposes. In the B. A. 

 Report for 1886 there is a provisional list of lines of the water- 

 spectrum, which he often uses, although the wave-lengths have 

 since been corrected. Further, if other lines are necessary, they 

 are found by halving the wavelengths of the secondary .spectrum 

 of hydrogen. Many lines thus determined are actually present 

 in the water-spectrum ; but why are not all there ? Dr. Griinwald 

 says it is because the amplitude of vibrations of parts of the mole- 

 cule can be so changed, owing to the presence of other substances, 

 that the intensity may increase or diminish, or become too faint 

 to be observed. To this argument there is absolutely no answer. 

 In some ca es, too, the average of two wavelengths is used as a 

 criterion of a wave-length of b which falls between them ! And 

 as a last resort, if the necessary wave-length cannot be found in 

 the water-spectrum by any of these means, it is put down as 

 'new,' and is called an 'unobserved' line. As just shown, 

 Dr. Griinwald ea-ily explains why the strongest lines in the 

 spectrum of an element, cadmium for example, when ' trans- 

 formed ' into water-lines, may be faint, and vice versii. But 

 how does he account for the fact that double lines are not trans 

 formed into double lines? This seems to me a fundamental 

 objection. The concave-grating gives the only accurate method 

 of determining the ultra-violet wave-lengths of the elements; 

 and, as a consequence of not using it, most of the tables of 

 wave-lengths so far published are not of much value. So 

 Griinwald's en or tiere may be great. And, besides, when we 

 consider that in the water-spectrum as given by Liveing ■ and 

 Dewar, without the "help of the secondary spectrum of hydrogen, 

 there is on the average one line for every two Angstrom units, 

 it would be remarkable indeed if any law could not be verified. 

 This is strikingly shown in the first group of the cadmium lines. 

 Here 6742 and 6740 are two readings for the wave-lengths of the 

 same hne, as made by two observers ; yet Griinwald finds a 

 water-line for each of them ! 



"The fact that there are exact numerical relations connecting 

 the spectra of different elements does not afford a proof of 

 Griinwald's hypathesis ; and until the above difficulties are re- 

 moved the evidence is against it. But, even granting it, how 

 do we know that a and b are not themselves compounds? In 

 the second group of cadmium lines there are nineteen hnes which 

 can be tr.'^n^formed into b lines ; b has many other lines ; so at 

 the most this only shows that cad nium and b have a common 

 constituent unless, of course, the absence of the other cadmium 

 lines is accounted for in Griinwald's own way of varying 

 intensity. 



"The lines of the spectrum of any substance, as carbon or iron, 

 seem to fall into definite series or gioups ; and the wave-lengths 

 of the lines in these groups can be expresse 1 by formulas, as is 

 well known. All that the fact of there being a connection 

 between the spectra of different substances seems to show is, 

 then, that there may be a formula common to many elements, 

 as Kayser and Runge have recently found. And all that this 

 means is that the molecules of those elements vibrate in general 

 according to a similar law." 



ON THE FORMATION OF MARINE BOILER 

 INCRUSTA TIONS.^ 



TN the older forms of marine boilers, sea waer uas almost 

 universally employed ; but with the intrjduction of high- 

 pressure tubular boilers t le amount of deposit was so serious, and 

 the difficulty of removing it so great, that it became inperative to 

 use distilled water. It is found, however, that the trouble has 



' A Piiper read at the thirtieth <!ession of the Institution of Naval Archi- 

 tects, by Pr<f. Vivian B. Lewis, K.C.S., F.l.C, Royal Naval College, on 

 April II, i8?9. 



