August 29, 1895] 



NATURE 



427 



reduced to oscillation-frequencies, and by means of Barlow's 

 mathematical tables these were reduced to wave-lengths which 

 are the reci])rocals of the oscillation-frequencies. The author 

 tlien went on tu describe some of the difficulties met with in 

 obtaining measurements of bands, due to alterations in wi<lth, 

 or to their becoming less distinct at the edges. The question 

 is dealt with in " Flanjc Spectra of High Temperatures,' Philo- 

 sophical Transactions, 1894, part I. 



Prof. Hartley had carried out experiments at Crewe, and 

 at Dowlais, in South Wales. Results obtained by photography 

 of the sjiectrum of the Bessemer flame were given in the 

 paper. For the details we must refer our readers to the original 

 memoir. As the author pointed out, the Bessemer spectrum is 

 a complex one, w-hich exhibits differences in constitution during 

 different periods of the blow, and even during different intervals 

 of the same period. Watts had observed that the spectrum 

 differs in different works, owing to variations of temperature and 

 the composition of the metal blown. After discussing the 

 various opinions held by previous investigators as to the utility 

 of spectrum analysis in steel making — on which subject inquirers 

 are by no means agreed — the cause of the non-appearance of 

 lines at the termination of the blow is discussed. Prof Hartley 

 then proceeded to what was jierhaps the most interesting part of 

 his paper, naTnely, the rempcrature of the Bessemer metal and of 

 the flame, antl the use of the spectrum as an index of tempera- 

 ture. Watts concluded that though the temperature of the 

 llame was above the melting point of gold, it was hcl'iw that of 

 platinum. Le Chatelier {Comptes renciits, vol. cxiv. p. 670) 

 was of opinion that the temperature of the Bessemer converter 

 during the boil is 1330 C, at the finish 1580° C, 

 while the steel in the ladle is at 1640" C. There is no 

 measure of the temperature .at the hottest period of the boil, and 

 unless the metal in the converter is cooled during the last minute 

 of the blow , which some of the author's photographs indicated, 

 it was difficult to understand how its temperature coidd be raised 

 l>y the addition of the cooler spiegeleisen and ferro-manganese. 

 The rise of temperature at this period could be accounted for by 

 the after-blow. Of course when the metal is charged with 

 'ixygen, the additional spiegeleisen, containing carbon and 

 manganese, woukl cause the combustion of these elements. 

 When the oxyhydrogen flame spectra of the manganese, magnetic 

 Mxide of iron, and ferric oxide are photographed, the number of 

 lines and bands in the spectra are not more numerous than with 

 a Bessemer flame spectrum of only half a minute's exposure, 

 although the above spectra may have received any exposure 

 from thirty to eighty minutes. When a substance emits a 

 spectrum composed of liands and lines, it is evidence of 

 ttie ])resence of the substance 'in the flame in a state 

 'if glowing vapour: when the same substance emits 

 two spectra, one differing from the other by the largely increased 

 number of bands or lines, it is evidence that either the suljstance 

 is more copiously vapourised, or that the temperature of the 

 vapour is higher. When a simple spectrum changes to one of a 

 more complex character, the alteration is due to an increase in 

 temperature, other things being equal. .Similarly when a 

 spectrum extends through the visible rays into the ultra-violet 

 region, and an increase is observed in the number and intensity 

 of the ultra-violet rays, nothing but an increase of temperature 

 will serve to account for the change in the spectrum. No 

 increase of material in the flame would increase the refrangi- 

 bility of the rays emitted by its vapour : hence the study of the 

 ultra-\iolet spectra of flames by the photographic method becomes 

 an important line of investigation. 



After pointing out the difficulty of ascertaining the maximum 

 temperature of any flame (as such temperature may exist over 

 but a very small area), and giving an instance, the author states 

 that Le Chatelier's recent measurements of the temperature of 

 furnaces have given numbers considerably lower than those 

 usually accepted. Langley's estimate of the temperature of the 

 Bessemer flame at 2000° C. — because platinum appears to 

 be rapidly melted in it —is not to be relied upon. Le Chatelier 

 finds that the metal is not fused but dissolveil in drops of 

 molten steel. Marshall Watts observed that the .sodium lines 

 5681 and 56S7 m.ay be enqiloyed as an index of temperature, 

 -Ince they are present in the spectrum of any flame containing 

 iilium the temperature of which is hot enough to melt platimnn, 

 liut they do not appear at lower temperatures. The Bessemer 

 flame does not show this double line, but only the D lines, 

 neither does it show lithium orange lines, which appear at a 

 ^'Muewhat lower temperature. It may therefore 'oe concluded 



NO. 1348, VOL. 52] 



that the flame is not hot enough to produce these lines. The 

 proportion of sodium in the Bessemer flame is evidently very 

 small from the narrowness and want of intensity of the D lines, 

 and the fact that they are not seen reversed in any spectrum ; 

 hence, though the temperature may be high enough, the quantity 

 of material present is not sufficiently large to yield the lines 5681 

 and 5687. 



We have not space to follow the author in all the interesting 

 details of his reasoning, but we have perhaps said enough to 

 indicate his line of thought. He later points out that, judging 

 by the number of lines and bands belonging to iron and 

 manganese, which have been photographed in the spectrum of 

 the Bessemer flame, the temperature must in any case nearly 

 approach that of the oxyhydrogen flame, even if it does not very- 

 generally exceed it. The paper concluded with particulars of 

 the heat of combustion of the oxidisable impurities in pig iron. 

 He calculates, as far as data are available, the absolute heating 

 effect of such oxidation. The temperature retained according 

 to these calculations amounts to 1454^ C. above that of molten 

 cast iron. This, however, is a theoretical value, and allowance 

 must be made for the specific heats of the gases, the metal, and 

 the slag, which are greater at the elevated temperatures than at the 

 temperatures at which the numbers representing specific heats 

 were determined. The specific heat of the converter must be 

 considerable, but it must be remembered that it is already heated 

 to the temperature of the molten metal ; but even if we allow 

 that 50 per cent, of the heat is absorbed, or conveyed away, we 

 should then have the temperature 727" C. above that of the 

 molten pig iron : and thus, with grey iron, at 1220' C. the metal 

 may have acquired a temperature of more than 1947" C. , which 

 is ver)' considerably above the melting point of platinum. 



The discussion which followed the reading of this paper was 

 interesting, but no new points of importance were added. Mr. 

 Bauerman considered that the author was right in laying stress 

 on the temperature of the flame as well as on the materials in 

 the converter. Mr. J. Stead pointed out that some of the cal- 

 culations were made in cases where the composition of the metal 

 was very different to that common in England. Mr. Tucker 

 also pointed out the difficulty in arriving at any conclusion owing 

 to the variation in metal used, and he referred to the effect of a 

 temperature of dissociation which might be obtained if the metal 

 were sufficiently rich in silicon. His own experiments supported 

 those of Prof. Hartley, that the temperature was certainly at 

 times considerably above the melting point of platinum, and he 

 was inclined to think that the temperature of dissociation was 

 often reached. 



The next paper was also one of considerable scientific interest. 

 It was Mr. Howe's contribution on the hardening of steel, and 

 was read in abstract by Mr. Brough, the Secretary- of the In- 

 stitute. As the paper had been received so recently, copies of 

 it had not been distributed, and it was manifestly impossible to 

 discuss a memoir of this alistruse nature at first sight, especially 

 as the paper was not read in full. It was therefore wisely deter- 

 mined to have the text corrected, after which the paper will be 

 distributed, and its discussion taken at the next meeting in May. 

 For the present, it will suffice to say that the .author deals largely 

 with the vexed problem of the allotropic state of iron. It woulil 

 have been a pity to have discussed the paper on the spot, as 

 neither Prof. Roberts- Austen nor Prof. .4rnold were present ; 

 neither had .M.- Osmond been able to send his usual written con- 

 tribution. In fact, the only person present whose name has be- 

 come at all prominently identified with the states of iron treated 

 was -Mr. Hadfield, who sixike briefly, saying that he had not 

 had time to master the paper. We will, therefore, defer 

 our abstract of this memoir until the time comes to give an 

 account of the next meeting. 



Mr. R. A. Hadfield's paper on the production of iron by a 

 new process was next read. The author's object has been to 

 obtain a pure iron ; for which purpose he had had recourse to 

 aluminium as an agent. The first result w.as that he made an 

 alloy of iron and aluminium very rich in the latter constituent, 

 there being no less th.an 36 per cent, present. In .spite of being 

 a failure, so far as the object in view was concerned, a very 

 interesting result was obtained ; for although there was no more 

 than a trace of carbon present, the alloy was hard enough to 

 scratch glass. Proceeding on the same lines, however, and 

 wcirking with ferrous oxide and granulated aluminium, a sample 

 of iron containing 9975 per cent, of that metal was finally 

 obtained at the very moderate cost of about eighteen pence per 

 pound. 



