98 Mr. W. N. Hartley .
On Origin of some the Lines Per cent , carbon .
The aspect of the spectrum .
1 *2 " 1 1*1/ 1-0 0-90 0-80 0 75 * 0-70 0-55 0-40 0-20 The group s just disappearing , the group 5 very distinct .
The group S a little fainter .
, , , , yet fainter .
S3 and \#163 ; 4 fainter than other bands of the group .
S3 and \#163 ; 4 indistinct .
S3 just disappeared .
\#163 ; 3 and 54 are away .
The group S away .
The group dx y3 ) p2 and generally y2 are away .
Subsequently no constant values were obtained , but on judging from similar spectra , on one occasion the metal contained 0'95 , and another time .065 per cent , of carbon .
The sole change that had .
been made was due to the slit of the instrument having been cleaned .
This indicates that it may be of some importance to have this so regulated by a micrometer-screw that its width may be accurately determined at any time during a series of observations .
The temperature of the iron of the converter may have some influence upon the disappearance of the bands , as similarly bands already described appear during the " blow " more distinctly if the temperature is high than if it is low .
" Remarks on the Origin of some of the Lines and Bands observed in the Spectra from Swedish Bessemer Works .
" By W. N. Hartley , F.R.S. Received April 1 , \#151 ; Read June 20 , 1895 .
The observations made at various Bessemer steel works in Sweden , as described in the preceding paper , are of particular interest , owing to the number of accurate measurements of bands and lines observed in different spectra of the flame during the progress of the " blow " when different varieties of metal are operated upon .
As the paper was written , as I have been informed by Mr. Lund-strom , about nine years ago , the author was at a disadvantage in being unable to compare all his measurements with flame spectra of known origin at high temperatures .
Fallacious deductions are liable to be drawn by referring bands and lines in the Bessemer flame to only such spectra as have been obtained by the arc or spark .
This was found to be the case in the study of oxyhydrogen blow-pipe spectra ( ' Phil , Trans , , ' vol. 185 , A , p. 161 ) ; hence I propose to offer \#166 ; 1 and Bands in the Spectra from Swedish Bessemer Works .
99 suggestions as to the origin of some of the bands and lines measured by Mr. Lundstrom .
The NyTcroppa Spectrum .
Remarks .
6079 is probably 6075 CaO 6028 , , 6026 Na Na Fe Fe K C l These have all been observed in oxyhydrogen flame spectra .
\gt ; Three bands in the green have been referred to phosphorus ; but it is very difficult to admit that this element , in any form , can give rise to their appearance in the Bessemer flame .
I have shown that phosphorus pentoxide yields a spectrum of its own , and is not dissociated in the oxyhydrogen flame ; also that ferrous phosphate yields the bands of phosphorus pentoxide .
It is just possible that when the blast is charged with moisture a compound cf phosphorus and hydrogen may be formed in the presence of much carbon ; but , on the other hand , such compounds show the green bands of phosphorus only when the flame is cooled , and , as that which issues from the mouth of the converter is intensely hot , the bands are scarcely likely to be visible .
In parallel columns are here given\#151 ; I. Measurements of the bands of phosphorus .
II .
Measurements of bands iu the Bessemer flame spectrum of Langshyttan .
III .
Bands to which it is suggested the Bessemer flame spectrum should be referred .
I. Phosphorus .
\ .
II .
HI .
Bessemer .
Reference bands .
\ .
X. 5581\#151 ; 5608 Probably 5608 Mn .
5264 " 5270 Mn .
5103 " 5110 Fe .
The Spectrum of the " Blue Flame " Several of the lines and bands measured in the spectrum of the " blue flame " are easily reproduced in the oxyhydrogen flame , others , however , which have been observed in the spectra which I have photographed at the Crewe Works of the London and North-Western 100 Origin of Lines and Bands Spectra from Bessemer Works .
Railway cannot with certainty be assigned to any particular element .
The following are some of the lines and bands referred to :\#151 ; Spectrum of the " blue flame .
" A. A. Remarks .
5709 is probably 5712 Mn .
5456 Mb .
5452 Crewe , 5338 Mn .
4983-5 Na .
4853 Mn from Mn02 .
4808'2 Crewe spectrum .
Li ?
4749-5 Mn .
4656 Mn .
4540 Crewe spectrum .
Fe .
4493 Crewe spectrum .
Fe .
Certain line 's or bands in the spectrum of the " blue flame " were found to coincide with , or closely approximate some of the principal bands measured by Salet in the spectrum of silicon , namely , 567 , 545 , 522 , 501 , and 487-5 .
A measurement corresponding to the first of these , described as 5672 , edge of band intense , was observed by me on plate 8 , Crewe , in the first spectrum only ; another measurement of the same band on plate 6 , Crewe , gave 5670 .
The origin of this band I have been unable to ascertain .
Of the remaining lines in the " blue flame " :\#151 ; A. A. Remarks .
545 is probably 5445 Mn .
Two lines , 5455 and 5452 , in the Crewe spectrum approximate to 5454"7 , which , according to Fievez and Thalen , is Fe .
522 " 5227 Mn203 .
Lundstrom .
Other lines corresponding to this are 5231 Crewe , 3rd spectrum , plate 6 , and 5217 Crewe , 3rd spectrum , plate 8 .
There is an iron line at 5231 .
501 " 5018 Mu203 .
487-5 , , 4862 Band coincident with the solar line F. It is , I believe , much more probable that silicon hydride is the cause of the bands in the spectrum , than that phosphorus is the origin of the three bands in the Langshyttan spectrum .
Neither silica , a silicate , nor a silicide , such as silico-spiegel , yields bands or lines of silicon in the oxyhydrogen flame , but a highly siliceous pig-iron , charged with much carbon , may react with water-vapour contained in the blast during damp weather so as to evolve carbon monoxide and the compound silicon hydride .
This , of course , On the Variable Stars of the 8 Class .
101 would readily burn in the air , and , while exhibiting the silicon bands it would also display the lines of hydrogen , particularly the C , F , and H lines , the red C being the most conspicuous .
Further observations on this spectrum are desirable , and the hydrogen lines should be particularly looked for ; but on the whole I am inclined to believe that Mr Lundstrom 's deduction as to the origin of the " blue flame being caused by silicon is substantially correct , if it be understood that the silicon is in combination with hydrogen .
Many of the lines and bands in the Ulfshytte spectrum , which have not been identified , have been observed in the Crewe spectra ; but their origin still remaius a mystery .
" On the Variable Stars of the 8 Cephei Class .
" By J. Norman Looker , C.B. , F.R.S. Received November 9 , \#151 ; Read November 21 , 1895 .
Introductory .
Professor E. C. Pickering , in his classification of the variable stars , which is based on a study of the light curves , * recognises two classes of variables having short periods .
His Class IV includes those variables , exemplified by 8 Cepbei and / 3 Lyrse , in which the light changes are not of very great range , and continue throughout the period .
Class V comprises those like Algol in which there is a\gt ; temporary reduction of light at .
minimum , produced by the eclipse of the bright star by a relatively dark companion ; this explanation has since been established by spectroscopic investigations , which have shown that there is no change in the spectrum at minimum , and that there is an orbital movement of corresponding period .
Both these classes are sharply distinguished from the variables of the IVlira type , which form Pickering 's Class II .
In this class of variables , as in Class IV , the light changes go on throughout , the period , but the range of variation is generally much greater , and the periods are reckoned by months instead of days .
The variability is associated in this case with a fluted spectrum ( Group II ) , and in a former communication I have shown that the phenomena can be explained on the supposition that there is one swarm ( or more ) of meteorites revolving in an elliptic orbit round a central swarm , an increase of light at maximum occurring at periastron when the number of collisions is greatest.f It is probable that the variables of Class IV will require further subdivision , when their spectra have been more fully investigated .
* ' Amer .
Acad. Proc. , ' vol. 16 , p. 17 .
t 'Roy .
Soc. Proc. , ' vol. 43 , p. 154 , 1887 .

