328 Capt. Abney and Gol .
Festing .
[ June 21 , resulted in some compound -which is a multiple of potassium ; potas-sium being , as we know , an important constituent of normal muscle .
Ba 137 x 2=274 274\#151 ; Ca 40=234 , K 39x6= 234 .
Rubidium in large doses has the same effect as barium in causing a veratria-like curve , but barium destroys the effect of rubidium before producing its own effect . .
Rb 85-4x8=683-2 , Ba 137x5=685 .
In the next division the authors show that by alternate application of acids and alkalies the muscle of the frog may be made to describe on a slowly-revolving cylinder curves which almost exactly resemble those described on a quick cylinder by the normal contraction of a muscle on stimulation ; and also those which the muscle describes on irritation after it has been poisoned by barium .
They consider that the contraction of muscle may be possibly due in some measure at least to alterations in acid or neutral salts which the muscle contains^ The lethal activity on frogs of the chlorides experimented upon is as follows : the potassium being most powerful , and calcium least powerful ; potassium , beryllium , rubidium , barium , ammonium , caesium , lithium , lanthanum , didymium , erbium , strontium , yttrium , , , sodium , calcium .
VII .
" The Influence of Water in the Atmosphere on the Solar Spectrum and Solar Temperature .
" By Captain Abney , R.E. , F.R.S. , and Colonel Festing , R.E. Communicated at the request of the Committee on Solar Physics .
Received June 14 , 1883 .
In our paper on " Atmospheric Absorption in the Infra-red of the* Solar Spectrum " \#166 ; ( " Proc. Roy .
Soc. , " vol. 35 , p. 80 ) , we stated that the absorption by water coincided with the absorption bands to be found in the solar spectrum , and our proof rested on photographs which had been taken for some time back .
In the diagram which we published ( and in which are slight errors in shading at some parts , and which we here correct ) we showed the coincidences as far as A.10,000 , that being the limit to which we could accurately fix the wave-lengths .
Simultaneously almost with the publication of our paper , there came into our possession an account ' of Professor Bangley 's researches on the solar spectrum , conducted by means of the bolometer , a perusal of which determined us to vary 1883 .
] On the Solar Spectrum and Solar Temperature .
329 _ raethod of observations in order to definitively test our conclu-.nns as regards intensity of absorption in the part of the spectrum below the 'red which we had explored , and also to ascertain if our deductions held good in the parts we had not explored .
It thus became necessary to go over the work done by Langley as far as our instrumental means would allow .
One of the most remarkable features in it seemed to be the failure of Cauchy 's formula for refraction , his prismatic spectrum extending below A to a distance equal to AG , whilst theoretically it should end at a distance equal to AE Our first labour was to ascertain the correctness of this by means of photographs taken with the diffraction grating .
Since 1880 , when the temporary map of the normal spectrum of the infra-red from \7000 to 10,500 was published , a large number of photographs for the determination of lower wave-lengths have been taken , with the result that a map as far as 15,000 can now be constructed , showing all the delicate Eraunhofer lines which exist in this lower region , but this extent is not sufficient for a comparison with Langley 's solar thermogram , which extends to \28,000 .
Unfortunately in our climate the atmospheric conditions preclude getting any results much lower than 15,000 , except on very rare occasions .
Advantage has been taken of cold* dry days to take rougher photographs , which though not so defined in detail as those up to A.15,000 , are yet sufficiently accurate to compare with Langley 's map .
His map is accurate to figures over the 100 on the tenth-metre scale , and our photographs have the same accuracy .
As a result of the comparison of our photographs to \22,000 , with the map , we may say , that the wave-lengths agree , and the failure of Cauchy 's formula is confirmed .
Our photographs were secured by separating vertically the different orders by means of a white glass prism of 30 ' .
The coincidences of the different spectra were thus readily seen , and the value at once ascertained .
The glass employed throughout our researches is the same as that used by Langley , to whom one of us recommended it , and of which he speaks in such high praise .
The use of rock-salt for anything except for purely absolute quantitative work becomes unnecessary , and for comparative work the glass is as effective and more certain .
Ho rock-salt prism which we have tried will bear the use of a collimator , the lens of which is filled by the radiation ; the surfaces have never been nearly as perfect as those of the glass .
The principal Fraunhofer lines were almost indistinguishable , unless the aperture of the collimating lens was very largely diminished .
This was the case with four separate prisms , which when first tried were newly polished by an optician .
Professor Langley has been more fortunate than ourselves in this respect , and has been able to compare his glass with rock-salt .
We owe , however , a debt of gratitude to Dr. Guthrie and Professor Tyndall for having given us rock-salt with which to experiment , and 530- Capt. Abney and Col. Festing .
[ June 21 -only wish our report of its behaviour might have been more favourable .
Having confirmed the correctness of Langley 's wave-lengths as fat as A.22,000 , we proceeded by means of the thermopile to examine the spectrum obtained when using the crater of the positive pole of the electric light as a source of radiation , and also subsequently the variations caused by placing different thicknesses of water in front of the slit .
The thermopile we used was constructed specially for us by Elliott Brothers , some two years ago , when we had made a commencement of this work.* It is a linear pile of twenty-six couples to the inch , and one face is covered by a silvered slit with moveable jaws .
The \#166 ; couples are mounted in the usual manner , and enclosed in a chamber with transparent ends which can be exhausted , if necessary .
Outside this again is a water jacket , through which a constant stream of cold water can be made to circulate .
The instrument is extremely delicate , \#166 ; and answered our purpose well .
Two reflecting galvanometers were j used at different times , one having an internal resistance of about j D'5 ohm , and the other of 0*12 ohm .
The thermopile was mounted !
on a stand , moveable horizontally by a screw of 18^ turns to the inch .
The range of movement was 6 inches .
The pile was always moved in one direction during one set of observations .
To form the spectrum the following arrangement was adopted .
A condenser of white glass threw an image of the crater on the slit of a collimator having a focal length of 20 inches , the lens of the collimator had a diameter of I4 inches , and the rays of light were rendered parallel for C , and then fell on a glass prism ( sometimes two ) placed at approximately the angle of minimum deviation for that ray .
A lens of 20 inches focal length attached to a camera formed the spectrum , which could be received either on a sensitive plate or on the face of the pile .
The width of the collimator slit was -S*5- of an inch for the electric lights , and for the sun inch .
The width of the slit of the thermopile was in the former case fainch , and in the latter fa .
In order to obtain fiducial points in the spectrum , the arc was focussed on the slit , a sodium spectrum formed , and a photograph taken .
The sensitive plate was replaced by the thermopile , and the D line brought in the centre of the jaws of the silvered slit , and this point on the screw was used as the zero of the scale .
From measurements of the photographs the exact position of all lines and bands could be readily obtained and referred to the reading taken with the thermopile .
A check photograph and zero reading were taken at the end of each series of observations .
An assistant kept the image of the crater on the slit ( which had a length of '31 inch ) , one of us attended to the movement of the pile , while the other took the galvanometer readings .
The unused face of the pile was covered with thick folds of non-con-* " Nature , " vol. 25 , 1881-82 .
1883.J On the Solar Spectrum and Solar Temperature .
331 ducting material , and the jacket of the pile was kept full of water , 'he whole being enclosed to form part of the camera by means if a thick covering .
At each movement of the pile at least three readings of the galvanometer were taken .
In each series of observations , and for each water spectrum examined , at least six series were made .
t \#166 ; We believe that the results we put forward are fairly representative Quantitatively , and without doubt the positions of maxima and minima .may be relied on to one-eighth turn of the screw. .
At the same time it must be remembered that the breadth of the slit of the thermopile is a measurable quantity , and one which cannot be neglected , but its jexistence is a necessary evil .
For this reason the positions of bands are more exact in the photographs than in the thermograms , and any small discrepancy in position should be discounted in favour of the photograph .
When using the solar radiation , which is four times more intense than that of the crater of the electric light , and when both slits have been narrowed , the absorptions are sure to be more intensely marked than when a wider slit is used , the reason of which is obvious .
Langley employed a spectrum which was of about the .same purity as ours ( H American Journal of Science , vol. 25 , p. Ib9 , 1883 , and " Phil. Mag. , " vol. 15 , p. 153 , 1883 ) , but the width of his bolometer strip being about one-fifth that of the jaws of the thermopile slit , and his spectrum longer , the absorptions are shown with greater strength than they are in our curves , whilst in our photographs the absorptions are even more marked .
Our first result was the thermogram of the crater of the positive pole , and this is shown in Curve I , Diagram I. This is the mean of a dozen curves which all lie very closely on this mean .
One thing we learn from it is that the temperature of the crater is always very uniform with any carbon , and is what we might expect when we consider that it is the temperature at which carbon vapprises .
In comparing it with Tyndall 's curve given in u Heat as a Mode of Motion , " we are struck by the difference exhibited in the two near the apices ; this difference , however , is explicable as will be seen further on , when it is remembered that his source of radiation was one composed of various temperatures .
Next , in order to ascertain the amount of absorption due to different thicknesses of water , it was found necessary to employ cells with glass sides , and , unfortunately , not being in possession of parallel glass of the same material as the prisms , it became necessary to take the absorption curve of the cells .
This is found in Curve II .
Water was then placed in the cells before the slit , and the readings taken .
Curves " VI , VII , and VIII are the water curves corrected for the cell-absorption , and they were constructed to enable a comparison to be made with the curve of the solar spectrum .
332 Capt. Abney and Gol .
Festing .
[ June 2 !
Annexed are tables showing the ordinates of the different points i the spectra as determined by the thermopile .
Curves# ( See Diagram I. ) Scale No. I. II .
III .
IY .
Y. YI .
YII .
YIII .
-10 .5 *4 #4 *4 0 *5 *5 0 - 8 1-3 .9 .9 .9 0 1*3 1*3 0 - 6 2*5 1-3 1*3 1-3 0 2*5 2-5 0 - 4 4*4 2-5 2-5 2-3 .2 4 *4 4-1 *3 - 2 5*9 4 *5 4 3 4*1 1*2 5*5 5-0 2*0 0 no 7*5 7 3 6-8 3*2 10 *5 9*5 4*7 This point is D. 12 -3 8-7 8*7 8-5 4 0 12 *3 12 *0 5-7 1 13 -7 10 0 9-2 8-8 3*7 13 0 12-3 5*3 li 15 1 11 -3 10-7 10 -o 5*0 14*0 13-3 2 16-8 12 8 12-2 11-6 6*7 16 *0 15 0 8 21 *5 16 0 15 9 15 *0 10*3 21 *4 20 0 Si 23 -0 16 -7 16-7 16 0 10 *7 23 0 22-5 15*3 H 24-2 17 5 17 5 16 7 8*5 24 *2 23 0 ' 4 27-5 19 *3 18 -5 17 0 2*3 26 0 24*0 3*5 This point is 1 4\#163 ; 31-5 21 1 21 0 19 -2 1 *8 31*5 28 *0 A. 5 35-0 22 *8 22 -8 20 -6 1*4 35 *0 32*2 2-0 5\#163 ; 38*2 24 6 23 0 21 -0 __ .
36 0 33 0 5\#163 ; 40-0 25 *4 24 -7 22-0 38*7 34 *5 6 41 -2 26 2 24 3 20-0 0 38*0 27*0 0 6i 42 3 27 0 23 *0 14 0 TXt \#151 ; \#151 ; OQ 6i 43 3 27-7 21 -5 5-3 0 33*7 8-5 CD O 6\#163 ; 44-5 28-4 24 0 6-0 e-t- \#151 ; \#151 ; 3 l g cL m 7 45 3 29-3 26 0 9 0 \#163 ; \#151 ; \#151 ; 7i 46-5 30 -0 26:3 10-0 P g m 40*2 15 *5 n 45 *7 29 4 25 0 9-7 \#151 ; .\#151 ; 8 43 2 27 -6 20-5 1*5 32 *2 \#151 ; The minimum 8i \#151 ; \#151 ; 18 -5 *5 28 *0 -8 probably lies a little 8| 40-0 25 -8 18 5 *5 \#151 ; \#151 ; , 8\#163 ; \#151 ; \#151 ; 19 0 .7 \#151 ; \#151 ; .
closer to 8 9 37 3 24-0 16 -0 .5 \#151 ; \#151 ; than 8| .
j 9\#163 ; 34-3 21 -4 6 0 0 9*7 0 10 31 3 18-5 .7 0 1*3 GQ *| O 10J 27 3 15 0 1*5 m V \#151 ; 11 24-0 11 6 3 0 a\gt ; a 5 *3 c-J- 3 1 m \#151 ; \#151 ; 1 6 rrf~ 3 | \#151 ; 12 15 -5 6-8 10 \#151 ; S sL 12\#163 ; \#151 ; \#151 ; 0 g P-\gt ; m 0 i 13 9-0 4-0 *4 \#151 ; QQ 13* \#151 ; \#151 ; 1-5 2*8 14 3-7 1-8 1 -1 \#151 ; 15 1 -8 .8 0 0 18 0 0 J A reference to the diagram will show that the thermograms of the water spectra have absolute coincidence with the absorptions shown in the photographs .
The thermogram is , however , more delicate in some respects ; as for instance , in the dip in the curves commencing at ^ in the scale .
This is a water-band which is only shown photographically and markedly with 4 feet of water* whilst with but L883 .
] On the Solar Spectrum and Solar Temperature .
333 1 inch of water it is distinct in the thermogram .
The reason of this f that when high intensities of radiation are nearly equal , an apparent equality of intensity of image is produced in a photograph when exposure is prolonged .
By shortening the exposure and thus anrrificinsr the rest of the spectrum , this band can be brought out .
Bussell and Lapraik ( see " Journal of the Chemical Society , " 1881 ) found this band visually when using a thickness of 14 feet of water , Apparently , therefore , the thermopile is more delicate for gentle shades of radiation than is the eye .
Diagram II shows the Curves I , VI , VII , and VIII , transformed to the normal scale of wave-lengths ( figured Ir , VI , VII , and VIII respectively ) , from which it will be seen that the maximum energy of the crater on this scale corresponds to a wave-length of about 7350 , or just at the limit of the red as usually visible .
It will also be seen that the depressions are not so marked as in the prismatic scale .
Diagram III gives the solar spectrum on two separate days .
The continuous curve was taken with the wet and dry bulb of the thermometer standing at 68 ' and 78 ' h. respectively .
According to usually accepted methods of calculation , this would correspond to an equiva883.1 On the Solar Spectrum and Solar Temperature .
335- ant 0f about 5^ inches of water between the sun and the spectro-oope .
The dotted curve was taken on a day which was drier , but n which the sky was a little hazy .
The following is a table of the ordinates of the continuous curve of jram III .
Scale No. Headings .
Scale No. Headings .
0 18 3 *5 ... .
35 0"2 . .
* * .
\#171 ; , ... 20 3-75 ... .
... .
38*5 1 26 4 ... - .
34 li 28 *3 41 34 x2 2 31 41 41 36#5 j-2 4f ... .
40 " 2 2# 37 5 ... .
34 *5 3 35 *5 54 ... .
... .
29*5 3-3 33 5 *5 ... .
... .
29 11883 .
] On the Solar Spectrum and Solar Temperature .
337 Scale No. Headings .
Scale No. Headings .
5*75 34 9*2 . . .
7 19 10 2 o3 . . . .
18 10i ... 22 of . . . .
\gt ; 7 28 11 25 # . . . .
38 in ... 16 ' i hi 35 12-| ... 0-5 / f \#151 ; 7i ... .
30 13t\\gt ; ... 7 8 20 131 ... ... . .
4 *5 8| ... .
19 14ro ... ... . .
8 0 2 25 141 0 0-5 . . . .
Qs 28 15 1 'T 9 21 -5 16 0 Diagram IY shows the continuous curve of Diagram III transferred to the normal scale ; and that in dotted line is Langley s -curve ( " Phil. Mag. , " March , 1883 ) transferred to our scale , so that the Tnq .
Yima .
correspond .
The chain-dotted curve is curve No. YI , Diagram II ; hut it must be borne in mind that the vertical scales are different , inasmuch as the slits of the thermopile and collimator were narrower in the case of the sun than in the case of the electric light .
To make an absolute comparison , the ordinates of .
the solar curve must be multiplied by 4*5 .
It will be seen that , generally , our curve corresponds with Langley 's , the difference in position of the maxima and minima being in no case greater than one-sixth turn of the screw of the thermopile .
His radiation receiver was so much narrower than ours , that his readings would show the rises and depressions more exactly .
A comparison of the curves will show that almost all the atmospheric absorption is due to watery stuff .
The depression shown in Langley 's and our own solar curves at 8660 is not coincident with water .
His shows the water-band at 8240 ( which is that shown in the photographs ) , and , no doubt , we missed it owing to the steepness of the descent from \7900 .
The dip at 8660 is probably due to a hydrocarbon , to which we will not here refer further in the present communication .
We wish to emphasise the fact that the " a ' group , corresponding to A.7200 , is not due to water as a liquid ; nor when we look at Diagram I and see the absorption of 2 feet of water , can we suppose that in the dip from 7300 to 7600 the A line is included .
Professor Langley in his paper states that the atmospheric absorption in the infra-red of the spectrum is comparatively small , and he gives a hypothetical thermogram of the extra-atmospheric solar spectrum in which the maximum energy is in the yellow , and the energy curve descends steadily on both sides of this maximum .
We do not think our experiments quite confirm his views in this respect .
" " 'VOL .
xxxv .
z 338 Capt. Abney and Col. Festing .
[ June 21 " We submit that a comparison of the thermograms of the sola ?
spectrum and those of the electric light points to the conclusion that much the greater portion of the absorption in the infra-red of the former is due to the presence of water-stuff .
In the case of the electric light it will be seen how large a proportion of the energy ig absorbed by even If inches of water , especially between \10,000 and A21,000 .
If water in the state of vapour , as we believe to be the case , absorbs as much energy as the same amount in a liquid state the sun 's rays are absorbed to at least a corresponding extent in passing through our atmosphere , which must , we believe , always contain more than an equivalent of If inches of water .
Now Langley 's normal solar curve may be considered representative of a thermogram taken under the most favourable conditions for dryness of atmosphere , and it cannot be supposed , even under-these circumstances , that between the spectroscope and the solar atmosphere there could have been less than |-inch of water ( or its ) equivalent in vapour ) , judging by the depression in it .
Fig. 2 , 'Diagram IV , shows the general form of his curve corrected for the !
absorption due to \#163 ; inch of water ; from which it appears that there must be a large proportion of energy emanating from a source of comparatively low temperature .
The prismatic thermogram of the spectrum from a source having a 883 .
] On the Solar Spectrum and Solar Temperature .
339 ingle temperature appears to be one having a sharply-defined maxi-nmn ( see Diagram Y ) .
Curve A is the curve of the crater of the lectric light , Curve B is that of an incandescent light produced with hirty-eight Grove 's cells , and Curve C that produced by thirty drove 's cells ; the same lamp being used in both cases .
The current ised was 2'25 amperes and the electromotive force 5T volts in the ast case , and in the former 2'95 amperes and 6'3 volts .
( The deflec-ions of the galvanometer attached to the thermopile , due to the naked ights , in the two cases were as 105 to 207 .
) The thermogram from i source of mixed temperatures would be the integration of those rom each temperature separately .
In illustration of this we have aken the three thermograms just mentioned , and transformed them 'see Diagram YI ) to the normal scale of wave-lengths , the ordinates \gt ; f B being exaggerated four times , and those of C eight times .
Curve D is formed by making the ordinates the mean between those of A md B , and the ordinates of E are the sums of those of D and C. Thus E would represent the thermogram of a source composed of the temperature of the positive pole of an arc lamp , an incandescent lamp worked by thirty-eight cells and one worked by thirty cells , in she proportion of 1 , 4 , and 16 .
In these last two curves , there is a ump , so to speak , which corresponds approximately in position to that shown in Langley 's solar curve as corrected for \#163 ; inch of water ( see tig .
2 , Diagram IY ) , indicating the presence of energy due to a low temperature source or sources .
The mottling on the sun 's disk when seen in a telescope makes it evident that at its visible surface there is a considerable range of temperature , and we have evidence of eclipse observations that outside its ordinarily visible surface there is much matter at a still lower temperature which is competent to produce a continuous solar spectrum .
With these amongst other facts before us , we are disposed to doubt whether the extra-atmospheric spectrum of the sun would give such a simple thermogram as Professor Langley has suggested , and whether the maximum energy would be found in the position in which he places it .
We believe that the curve would have a form rather more allied to our Curve E. At first sight it might appear hopeless to try and affix an approximate value to any solar temperatures , but we think that it is not impossible to determine with some approach to probability the maximum temperature to which any compound curve is partially due , more particularly when , as in the case before us , the general form of the corrected curve indicates an excess in quantity of low temperature .
Prom an inspection of the curves in Diagram Y it appears that for any temperature higher than that of Curve A the position of the maximum will be but very slightly shifted towards the more refrangible eiM of the spectrum , also that the general form of the curve must be similar to that of A , and that the areas within the curves , which 1883 .
] On the Solar Spectrum and Solar Temperature .
341 are measures of energy , will be very nearly proportional to the ordinates of the maximum or to ordinates not far from them on the [ more refrangible side .
Now the part of the spectrum which suffers least absorption by water , and which in the solar spectrum is free from any very intense Fraunhofer lines , is near 4 on the prismatic scale .
It has been shown by ; Dewar , * and be deduced the same from Rosetti 's formula , that the \#166 ; temperature of the source is nearly proportional to the square root of ; the total radiation , or , in other words , of the area of the thermogram | curve .
The incandescent lamp worked by thirty cells had approximately a temperature of 1,100 ' , and the same when worked by thirty-eight cells of 1,500 ' , and the areas of the curves in the two cases are as 106'4 to 50-4 , and the galvanometer deflections when compared together , as already stated , as 105 to 207 , or about two to one in both cases .
This would give about the temperatures above stated , taking either one as correct .
The area of the curve of the crater is very nearly sixteen times that of the thirty-eight cell curve , or the temperature of the crater , using Dewar 's formula , would be about 6,000 ' , a temperature corresponding to that obtained by him .
We have already shown at what point of the solar spectrum aqueous absorption has least effect , and if we may .
calculate from this the probable area of the curve of maximum temperature .
In our prismatic solar thermogram the scale of the ordinates is 4*5 times the scale of the crater thermogram , but the ordinates of the point of the spectrum above referred to are nearly equal .
Hence the highest solar temperature would be ^4'5 x 6000 ' , or about 12,700 ' , using the same relations of temperature and radiation.f Taking , however , the square root of the whole area of the solar curve , as would be the case when direct and not spectrum measurements were made , we should get a temperature of about v/ 2 '5 times that of the crater , or 9,600 ' , a temperature also very similar to that obtained by Dewar under the conditions specified .
In conclusion we say , 1st , that the thermopile experiments have confirmed our previous views as to the coincidence of the absorptions by water and those shown in the solar spectrum ; 2nd , that the extension of the work beyond our previous point emphasises them ; 3rd , that the highest temperature of the sun is not less than 12,700 ' , a temperature far higher than that which has been recently put forward by Sir W. Siemens ; 4th , that the existence of a large quantity of solar radiation due to low temperature has been shown to be more than probable .
* " Brit. Assoc. Rep. , " 1873 , p. 465 .
t Since this paper was read we have made further experiments in regard to this relation , and believe that we may have to modify it , reducing the temperature somewhat.\#151 ; July 29 .

