509 The Fluted Spectrum of Titanium Oxide .
By A. Fowler , A.R.C.S. , F.R.A.S. , Assistant Professor of Physics , Royal College of Science , South Kensington .
( Communicated by H. L. Callendar , M.A. , LL. D. , F.R.S. , Professor of Physics , Royal College of Science , S.W. Received May 18 , \#151 ; Read June 20 , 1907 .
) [ Plate 6 .
] In a previous paper* it was shown that most of the dark flutings which are characteristic of the spectra of Antarian or third type stars correspond with flutings which appear in spectra obtained from oxide and chloride of titanium , but it was then uncertain whether the flutings originated in the vapour of the oxide or in that of the metal itself .
The purpose of the present communication is to give an account of the observations which have led to the conclusion that the flutings in question are produced by a compound of titanium with oxygen , and to give a revision of the table of wave-lengths based upon photographs taken with increased dispersion .
The Oxide Origin of the Flutings .
The general principle underlying the investigation was the very simple one that if the " Antarian " flutings were due to the element titanium they should appear , under favourable conditions , in the spectra of compounds of the metal with different elements , whereas if they were produced by the oxide they would only appear when the metal was combined with oxygen , or was able to combine with it during the experiment .
The most decisive results have been obtained from experiments with titanium chloride ( TiCl4 ) , a heavy volatile liquid , the vapour of which rapidly combines with oxygen on exposure to moi\amp ; t air and forms dense fumes of titanium oxychloride .
On account of this affinity for oxygen , considerable care is necessary in attempting to obtain the spectrum of the chloride without contamination with that of the oxide .
The observations have shown , however , that not only is the spectrum of the chloride free from the Antarian flutings , but that it is characterised by a perfectly different group of flutings in the blue which does not occur in the stellar spectra .
This " chloride group , " as it may be conveniently called for purposes of reference , is a somewhat complicated cluster of flutings fading towards the violet , having three principal heads at wave-lengths 4199*5 , 4192*7 , and 4188*0 , of which the middle one is the brightest .
* 'Roy .
Soc. Proc. , J vol. 73 , p. 219 , 1904 .
Mr. A. Fowler .
[ May 18 , Experiments with metallic titanium have also indicated that the Antarian flutings are only produced in the presence of oxygen .
Particulars of the experiments are as follows:\#151 ; ( 1 ) Vacuum tubes containing the vapour of titanium chloride were prepared in the usual manner , but it may be remarked that , in consequence of the low conductivity of the vapour , the most satisfactory results were obtained with tubes giving a total length of discharge not exceeding two inches .
When the capillary tube was very narrow , little more than a line spectrum was observed , even when there was no jar in the circuit there was just a trace of the chloride flutings already mentioned , and enhanced as well as are lines of titanium were well brought out .
In the spectrum of the bulb , however , the chloride group was a conspicuous feature , and enhanced lines were not seen , while arc lines were numerous .
When a tube of wider bore was chosen , the spectrum of the capillary corresponded very closely with that of the bulb in the previous experiment .
In each case , the passage of the jar spark resulted in the appearance of the line spectrum of chlorine and the almost total suppression of that of titanium .
No traces whatever of the Antarian flutings were obtained in these experiments , although the presence of the chloride group of flutings indicated that the electrical conditions in some cases were not unfavourable for their production , if their existence depended only upon the presence of titanium .
( 2 ) The spectrum of titanium chloride was further investigated by introducing some of the liquid into a tube containing dry nitrogen at atmospheric pressure .
A tube 2 cm .
in diameter and about 10 cm .
in length was sealed at one end and provided with platinum electrodes , having a sparking distance of about 1 cm .
Through a rubber stopper at the other end two smaller tubes were passed , one for the admission of any desired gas and the other for the escape of the gas driven out .
The admission tube was of ( - form , and attached to the vertical arm by a rubber connection was a previously prepared tube containing a small quantity of titanium chloride .
This receptacle consisted of a piece of glass tubing about 3 cm .
long , drawn out and sealed at its lower end , and scratched with a file so that it might be easily broken ; a short piece of rubber tubing was attached at the other end , and when a sufficient quantity of the liquid had been poured in to fill both glass and rubber tube , the latter was closed with a pinch-cock .
In this way a sample of the liquid practically free from oxide was secured , and it could be introduced into the sparking tube when desired without further exposure to the external air .
The apparatus having been thoroughly dried by a current of warm dry air , 1907 .
] The Fluted Spectrum of Titanium Oxide .
511 it was filled with nitrogen , and the titanium chloride was then admitted by breaking off the end of the tube containing it .
Even after all the precautions taken to exclude moisture , there was a very slight formation of fumes on the exposure of the liquid , which , however , subsided in a few minutes .
On passing the spark ( without jar ) numerous lines of titaniumland the chloride group of flutings were seen , but there was no trace of the Antarian series of flutings .
When the jar spark was passed , the chloride flutings were abolished , and enhanced lines of titanium were strongly marked .
To carry the experiment a stage further , the nitrogen was replaced by dry oxygen , and the trace of oxychloride fumes again resulting from residual moisture was allowed to subside .
On passing the spark the Antarian series of flutings formed the most prominent feature of the spectrum , but the chloride group was also visible .
( 3 ) A similar experiment in which the sparking tube was filled with dry hydrogen gave an identical result .
That is , the chloride group of flutings was visible , while the Antarian series was absent .
In this experiment a purple deposit , presumably of titanous chloride ( Ti2Cl6 ) , was formed on the walls of the tube .
( 4 ) Experiments on the arc spectrum of titanium chloride ( on iron poles ) in an atmosphere of dried nitrogen were also made .
Under these conditions , the line spectrum of titanium and the chloride group of flutings were well developed , but the Antarian flutings were excessively feeble and the traces observed were probably due to residual oxygen or moisture , as a thin white deposit was formed on the poles .
With an atmosphere of oxygen , however , the Antarian flutings came out strongly .
( 5 ) In another series of experiments , the arc was passed between iron poles , charged with metallic titanium , in an atmosphere of dried nitrogen .
Only feeble traces of the Antarian flutings were seen , and these may again be sufficiently accounted for by residual oxygen or moisture .
The flutings , however , were well developed when oxygen or air were substituted for nitrogen , but in neither case was the chloride group observed .
( 6 ) As already noted in the previous paper , the Antarian flutings are also visible under the following conditions:\#151 ; ( a ) Arc in air between carbon poles well charged with oxide of titanium , the flutings being best seen in the " flame " when the arc is long .
( b ) Spark , without jar , through fumes of oxychloride of titanium ; with a spark of suitable intensity , the line spectrum is almost eliminated , but there is a continuous spectrum of considerable strength which doubtless arises from particles which are incompletely volatilised .
Mr. A. Fowler .
[ May 18 , ( c ) Oxy-coal-gas flame , fed with the fumes of titanium oxychloride ; the flutings are in this case not easily seen on account of the bright continuous spectrum which is also present .
These observations do not seem to admit of any other conclusion than that the flutings represented in the spectra of Antarian stars are produced by a compound of titanium with oxygen , and not by the vapour of the metal itself .
The result is of some importance as indicating that the source of the fluted absorption in the Antarian stars is at a temperature low enough to permit the formation of a chemical compound , and also as demonstrating the presence of oxygen , of the existence of which in these stars there is otherwise no direct evidence .
The investigation has lately gained additional interest in consequence of Professor Hale 's discovery of some of the less refrangible flutings in the spectra of sun-spots.* The Wave-lengths of the Flutings .
The identification of the flutings of the Antarian stars with those of titanium rested upon such a great number of apparent coincidences in position , and similarity of appearance , that it was almost independent of a very precise knowledge of the wave-lengths , and was sufficiently justified by the wave-lengths derived from photographs taken with the moderate dispersion then available .
No further determinations of the positions of the stellar bands have been published , but the application of a more powerful spectrograph has made it possible to determine the wave-lengths of the terrestrial flutings with much greater precision .
The instrument employed was a very efficient one of the Littrow form , having one prism of 60 ' = 1*6467 ) and a focal length of 12 feet , the effective aperture employed being 1^ inches .
The spectrum is photographed in sections on plates 12 x 2\#163 ; inches , and the linear dispersion ranges from 16 tenth-metres per millimetre at A. 7100 to 2*7 tenth-metres per millimetre at A 4350 .
With this dispersion the heads of many of the flutings are found to be more complex than was formerly suspected , but neither this nor the corrections of the wave-lengths affects the probability of identity with the stellar flutings .
Many experiments have been made in order to produce the fluted spectrum as free as possible from superposed lines .
The uncondensed spark passed through the fumes from titanium chloride is possibly the best way of obtaining this result , but the photographic registration with high dispersion is difficult .
The most convenient method yet found is to volatilise titanium * ' Astrophysical Journal , ' vol. 25 , p. 75 , 1907 .
1907 .
] The Fluted Spectrum of Titanium Oxide .
513 oxide in the electric arc between iron poles , and the photographs reproduced in Plate 6 were obtained in this manner .
Under these conditions , the " line " spectrum of titanium is not strongly marked , and iron is only represented by the brighter lines of its flame spectrum .
Nearly all of these metallic lines appear on the more refrangible side of D , and are strongest towards the blue .
The flutings which occur in the flame of the iron arc do not appear to contribute appreciably to the combined spectrum of iron and titanium oxide .
Incidentally , the photographs admirably illustrate the simplicity of the line spectrum of iron in the arc-flame .
For the determination of wave-lengths , the reference lines employed were those of iron and titanium occurring with the flutings , whenever suitable lines were available for the purpose .
In many cases , however , and in the whole region on the red side of D , the reference lines were selected from a spectrum of the iron arc photographed in juxtaposition with the fluted spectrum ; the shutter for exposing the two parts of the slit was entirely detached from the spectrograph , and there was no evidence of relative shift of the two spectra in the plates measured .
The adopted positions of the reference lines were those given by Eowland in his table of solar spectrum wave-lengths , and the interpolation was made in the usual manner by the Cornu-Hartmann formula .
For the part of the spectrum less refrangible than 6860 , in which Eowland does not tabulate any lines of iron , lines of the arc spectrum were first identified with solar lines and the corresponding solar wave-lengths adopted .
The wave-lengths given in the table are stated to two places of decimals , except in the extreme red , where .the dispersion is relatively small , and in cases where the edges of the flutings are not very sharply defined .
No attempt has yet been made to tabulate the thousands of fine " structure lines " which compose the flutings , but all the " heads " and " sub-heads " which could be identified as such have been included .
Some of the more marked details in the heads , probably consisting of relatively strong structure lines , or ( dusters of such lines , have also been measured ; for want of a better term , they are described as " maxima .
" The classification into heads and sub-heads is somewhat arbitrary in many cases , and it should be explained that some of the heads are classed as such , not because they are prominent features of the spectrum , but because of their probable association in series with stronger heads in the groups to which they belong .
With the high dispersion employed , the heads of some of the fainter flutings , especially in the region more refrangible than F , almost lose their ?
distinctive characteristics , and their identification has only been possible in vol. lxxix.\#151 ; -Jl .
2 N Mr. A. Fowler .
[ May 18 , some cases by comparison with photographs taken on a smaller scale .
It has been considered desirable to include these for reference in comparisons with stellar spectra , which are usually photographed with moderate dispersion , and also because of their possible use in investigations of the series relationships of the various flutings .
The relative intensities of the flutings are roughly shown by the numbers in the second column of .the table , and an attempt is made in the third column to indicate the general characteristics by symbols having the following significance:\#151 ; ( a ) The fluting fades out rapidly , and is not clearly resolved with the dispersion employed .
These appear as lines with a slight shading towards the red .
( b ) The fluting fades away more gradually , and the structure lines are very closely crowded together .
( c ) The fluting resembles those of class ( \amp ; ) , except that the structure lines are more clearly separated .
( d ) The structure lines are widely separated , and extend over a long range .
For convenience of description in the table , the flutings are classed in numbered groups , into which the spectrum seems naturally to divide itself .
It is not possible , however , to convey an adequate idea of so complex a spectrum by means of a table alone , and reference should be made to the reproductions of the photographs given in Plate 6 .
The first four strips , representing the spectrum from 4580 to 7200 , have been enlarged 1*8 times from the original negatives , and the attached scale of wave-lengths will facilitate comparison with the table .
Lines due to iron are separately indicated , as also are those arising from impurities of sodium , calcium , and lithium in the material employed .
The superposition of iron lines on the flutings beginning at 4954*8 , 5167*0 , and 5448*5 , ' tends to conceal the character of the heads , and reproductions of photographs taken with carbon poles are accordingly given in the fifth strip , the enlargement here being 3*5 times .
In this case most of the iron lines are absent , but titanium lines are generally more numerous throughout the spectrum .
The photographs of the less refrangible parts of the spectrum were taken on Messrs. Wratten 's " Verichrome " and " Panchromatic " plates , which gave uniformly good results .
1907 .
] The Fluted Spectrum of Titanium Oxide .
The Fluted Spectrum of Titanium Oxide .
515- Wave- length .
Intensity .
Character .
Remarks .
4353 *68 2 d 1st head of 14th group .
4395 *05 2 d 2nd head .
4421 -66 1 d Sub-head .
4436 '68 2 d 3rd head .
4462 *34 3 d 1st head of 13th group .
4462 70 3 \#151 ; A " maximum * ' in head .
4506-08 2 d 2nd head .
4506 -62 2 \#151 ; A maximum in head .
4548 -04 2 d 3rd head .
4548 *30 2 \#151 ; A maximum in head .
4584 -62 3 d. 1st head of 12th group .
4584 -92 3 \#151 ; A maximum in head .
4586 -78 3 d Sub-head .
4587 *20 3 \#151 ; A maximum in sub-head .
4626 -49 4 d 2nd head .
4628 -68 4 d Sub-head .
4668 *82 4 d 3rd head .
4669 *19 4 \#151 ; A maximum in head .
4671 *26 3 d Sub-head .
4671*66 3 \#151 ; A maximum in sub-head .
4761 *08 5 d 1st head of 11th group .
4761 -37 5 \#151 ; A maximum in head .
4761 *86 5 - \#151 ; 4764 *52 5 d Sub-head .
4804 *55 4805 *61 } 5 d 2nd head , included in cluster .
4807 *42 4 d Sab-head .
4848 *20 4849 *03 } 3 d 3rd head , included in cluster .
4893 *00 2 d 4th head .
4954 *78 6 d 1st head of 10th group , sharply defined .
4955 *26 6 \#151 ; A maximum in head .
4957*21 6 d Sub-head .
5002 -88 5050 -5 3 3 d d 3rd head 9 } Identification difficult with high dispersion .
5167 -00 7 d 1st head of 9th group , sharply defined .
5167 -50 7 \#151 ; A maximum in head ( not Mg line ) .
5169 -51 7 d Sub-head .
5240 -71 5 d 2nd head , structure lines very wide apart .
5241 *00 5 \#151 ; A maximum in head .
5307 *92 3 d Sub-head .
5308 *14 3 \#151 ; A maximum .
5356 -21 2 d Sub-head .
5356 -86 2 \#151 ; A maximum .
5359 *07 3 \#151 ; A maximum , or ?
sub-head .
5359 *51 2 \#151 ; A maximum .
5361 -22 3 \#151 ; \#187 ; A maximum , or ?
sub-head .
5391 *05 2 \#151 ; 5448 -48 7 d 1st head of 8th group , sharply defined .
5449 *07 7 \#151 ; A maximum in head .
5451 *32 7 d Sub-head .
5496 *79 5 d 2nd head .
i .
Mr. A. Fowler .
[ May 18 Wave- length .
Intensity .
Character .
Remarks .
5597 -92 10 a 1st head of 7th group , sharply defined .
5603 -98 5 b Sub-head .
5629 *53 8 a 2nd head .
5635 *54 5 b Sub-head .
5661 *68 6 a 3rd head .
5667 -81 4 b Sub-head .
5694 -56 5 a ?
4th head .
5728 -13 4 a ?
5th head .
5760 '15 4 b 1st head of 6th group ( followed by Ti 62 *48 and Fe 63 *22 ) .
5790 '86 4 b Sub-head .
5811 *28 4 b 2nd head .
5815 *14 4 \#151 ; A maximum , or line .
5846 -70 4 b Sub-head .
5863 *55 4 b 3rd head ( probably ) .
5872 -9 3 b Sub-head .
5905 *1 3 b a 5954 '66 3 b 1st head of 5th group ( relatively inconspicuous ) .
6006 *5 3 b 2nd head .
6057 *6 2 b 3rd head .
6149 -2 5 a ?
A doubtful head .
6158 *86 10 a 1st head of 4th group .
6162 *37 8 \#151 ; Sub-head P Perhaps wholly Ca line .
6174 -60 10 a 2nd head .
6183 *83 4 b Sub-head .
6186 *77 8 a 3rd head .
.6190 *07 5 b Sub-head .
6215 *35 8 a 4th head .
6222 *72 6 b 5th head .
6268 *35 4 b Sub-head .
6275 *70 3 b 6321 *0 2 c 6321 -8 2 \#151 ; A maximum in sub-head .
6350 *6 2 b Sub-head .
6357 *9 3 * c 1st head of 3rd group ( relatively inconspicuous ) .
6384 -4 3 c 2nd head .
6400*1 2 b Sub-head .
6416 *0 3 b 3rd head .
6448 *2 4 b 4th head .
6479 *4 4 b 5th head .
6484 *0 4 Sub-head ?
or cluster of lines .
6512 *8 4 b 6th head .
6544 *5 4 c 7th head .
6550 *2 4 \#151 ; - A maximum .
6551 *8 4 \#151 ; 6562 *5 3 c Sub-head .
6579 *5 3 c 6594 *5 3 \#151 ; A maximum .
6626 *3 3 \#151 ; a 6634 *4 3 b Sub-head .
The Fluted Spectrum of Titanium Oxide .
Wave- length .
Intensity .
Character .
Remarks .
6651*5 8 b 1st head of 2nd group .
6681 *0 8 b 2nd head .
6714 -1 8 b 3rd head .
6748 *0 8 b 4th head .
6782 *0 8 c 5th head .
6815 *1 5 c 6th head .
6850 -0 5 i c 7th head .
6852 *5 6883 '8 5 4 J c 8th head .
6919 -4 4 c 9th head .
6951 -8 2 c 10th head .
6988 -8 2 .
c 11th head .
7054 -5 10 a 1st head of 1st group .
7059 -6 4 b Sub-head .
7087 -8 10 a 2nd head .
7092 9 4 b Sub-head .
7125 -5 10 a 3rd head .
7130 -6 4 b Sub-head .
7158 -9 3 a 4th head ?
7197 -7 2 a 5th head ?
Reference to Stellar Bands .
The use of greater dispersion has removed*the doubt as to the agreement of the 7th group of flutings with Duner 's stellar band No. 4 , to which reference was made in the former paper .
It is now evident that the appearances described as " lines " near 5598 , 5630 , and 5662 are really the heads of short flutings of class a , as might have been suspected from the near equality of the intervals separating them from the weaker but more obvious flutings ( when seen with small dispersion ) at 5604 , 5636 , and 5668 .
This group , therefore , begins with the fluting at 5597*9 , and there is no longer any uncertainty as to its general correspondence with the stellar band , for which Father Sidgreaves gives the wave-length 5597 .
Evidence as to the presence in stars of the first group of titanium oxide flutings , beginning at 7054*5 , is afforded by the recent photographs of the spectrum of Omicron Ceti taken by Mr. Slipher , of the Lowell Observatory.* It is stated that " the star spectrum stops so suddenly at A , 7040 as to leave little doubt that another of these bands begins at that point and outruns the sensitiveness of the plate into the red .
" There can be no doubt that the abrupt ending to which reference is made owes its origin to absorption by the 7054*5 group of titanium flutings .
Still more recently , Mr. Newall has informed me that he has identified the three heads of the 7054*5 group in photographs of the spectrum of a Orionis .
* 4 Astrophysical Journal,5 vol. 25 , p. 236 , April , 1907 .
518 The Fluted Spectrum of Titanium Oxide .
All the stronger groups of flutings have accordingly now been traced in stars , and it may be assumed that the fainter flutings of groups 3 , 5 , and 6 will also be present when the banded spectrum is well developed , as in a Herculis and o Ceti .
It still seems improbable , however , that titanium oxide can sufficiently account for all the stellar bands .
As remarked in the former paper , ,the stellar intensities of Dun\amp ; r 's bands 1 and 3 , beginning near 5862 and 6493 , appear to be too great to be wholly accounted for by the titanium oxide flutings which occur in these parts of the spectrum .
This is especially the case with the band at 5862 , which is very frequently a wide and dark band in the stars , whereas the fluting at 5863*5 is an inconspicuous feature of the titanium oxide spectrum .
The flutings in the neighbourhood of 6493 also seem to be too feebly marked in the titanium spectrum to produce a conspicuous absorption head in the stars .
More accurate wave-lengths of the stellar bands in question are urgently required for further investigations of their origin .
The author is anxious to express his indebtedness for assistance rendered at different times by students in training assisting in the Astrophysical Department of the Koyal College of Science .
Valuable aid in carrying out the experiments described in the paper was given by F. W. Jordan , B.Sc. , and J. Prescott , M.A. , and the large-scale photographs were taken by H. Shaw , A.R.C.S. , and E. J. Evans , B.Sc. The author is alone responsible for the determination of wave-lengths .
FLUTED SPECTRUM of TITANIUM OXIDE

