]\gt ; On Harmonic Ratios in the Spectra of .
On Harmonic Ratios in the Spectra.of Gases By ARTHUR SCHUSTER , Ph. D. , F.B.S. Received January 10 , 1881 .
It would be a matter of the greatest importance if we could disover an empirical law connecting together the diiferent periods of bration in which we know one and the same molecule to be capable swinging .
According to the most simple supposition the vibrations might be harmonical overtones of one fundamental note .
Varioustattempts have been made to prove that such indeed is the case , and the wave-lengths of diflerent spectral lines bear to each other the ' of two comparatively small integer numbers .
M. Lecoq de and Professor Johnstone Stony , especially , have disussed this question ; the wave-lengths used by the former do not possess the accuracy necessary for a final settlement of the point , but { Professor Stoney has , in the case of hydrogen , shown that three out of four lines in the visible part of the spectrum have wave-lengths , which , to a high degree of accuracy , are in the ratios of 20 : : 32 .
I have occupied myself at various times during the last ten years .
with this question , and have naturally accumulated a large quantity of material .
About three years ago , however , I came to the conclusion that only a systematic investigation could lead to a decisive result .
In any spectrum containing a large nnmber of lines , it is clear that , owing to accidental coincidences , we shall always be able to find ratios which agree very closely with the ratios of small integer numbers .
We can , however , by means of the theory of probability , calculate the number of such coincidences which we might expect to find on the supposition that no real law exists , and that all the lines are distributed at random throughout the whole range of the visible spectrum .
, on calculating ont all fractions which can be in a spectrum by any pair of lines , the number of ratios , agreeing within certain limits with ratios of integer numbers , greatly exceeds the most probable number , we should have reason to suppose that the lines are not distributed at random , but that the law suggested by Messrs. Lecoq de Boisbaudran and Stoney is a true one .
I have been engaged during the last three years in discussing some of the spectra in the manner indicated , and I wish to lay the results of the investigation before the Royal Society .
I took the spectra of the following elements ; the numbers in brackets indicate the number of lines for each body:\mdash ; bIagnesium . . . . . . . . . . . .
( 7 ) Sodium . . . . . . . . . . . . . .
Copper . . . . . . . . . . . . .
15 ) Barium . . . . . . . . . . . . . .
Iron . . . . . . . . . . . . . . . . . .
( 149 ) Similar ratios were for all possible fractions in the spectrum .
Now , if the lines in are distributed at random , we should expect the latio of the two differences to range criminately between and 5 ; the mean of all of them coming near hird talues obtained , which , entioned , oughtmean should bmaller tesult iable Iecond column gives tumbers oractions fpectrumwe should expect agreater number omall fence ther hmonic ratios itrue o to be near25 , if the lines are distributed at random .
Table I. Nothing could be more decisive against the law of harmonic ratios this table ; three out of the five elements considered , including the two containing the greatest number of lines , give a mean value greater than25 .
In order to see how near to this value we should expect the mean to ( come if no law connects the different lines , I have given the probable On Harmonic Ratios in the of Gases .
339 nviation from 25 in the fourtl ] column .
The term probable in pronble deviation is here used in the same sense as " " probable error has been calculated by means of the approximate formula\mdash ; erfc .
number of lines in each spectrum , is the probability that the mean value lies between ; for equal one-half , is the probable deviation .
It will be noticed that the actual deviation never differs much from probable one , but that it is greater for the two elements the number of lines .
If , therefore , any deduction is to be drawn the preceding table , it is that the ratios formed by two given es rather seem to harmonic ratios .
The method just explained , and which has given us such decidedly negative results , I believe to be very well adapted for the discussion spectra have a comparatively small number of lines ; but the iron spectrum may be examined by a more direct and complete n1ethod .
We may directly calculate how many fractions ought to agree within certain small limits with harmonic ratios if no law exists , and counting how many do thus coincide .
I have found , for instance , twenty-eight pairs of lines which coincide within limits so narrow that they can be easily due to errors of measurements with fractions , denominator and numerator of which are both smaller than 10 .
This number might appear large at first sight , and some support for the law of harmonic ratios might be derived from it .
But the caloulation gives the larger number 32 as the one we ought to expect , if all the lines were distributed at random ; so that here , also , the fractions seem to avoid rather the harmonic ratios .
A little difficulty is experienced in fixing the limits within which we may consider a coincidence to have taken place .
They must depend , of course , on the accuracy which we assign to ngstrom 's 1 measurements .
I thought it best to work out the results with two different limits , one of which was half as large again as the other .
We gain a decided advantage in classifying the results for two limits .
It is in fact equivalent to using a third of discussion , for supposing the spectral lines to be distributed at random , the number of coincidences found should be proportioned to the limits chosen .
If , on the hand , the law of harmonic ratios is correct , the narrower limit should relatively show the greater number of cidences .
The limits taken were\mdash ; and At first sight the result seems again decidedly against the theory of harmonlc ratios .
For all fractions with denominator smalIer than 70 the calculated coincidences are in excess of the observed ones .
seems , however , to be a greater number of ratios than we should expect , which agree nearly with fractions , the denominators of which lie between 70 and 100 .
If we compare the results given for the two different limits wo find that the smaller limit gives results decidedly more favourable to the theory than the larger ones , and that , as has been explained , is an lmportant fact which cannot be lefb out of account .
In the following Table , I have compared the number of coincidences for the limit with those calculated from the larger one , on the supposition that the coincidences are proportional to the limits , as they ought to be if no connexion exists between different lines of the samo spectrum .
It be seen , that with the exception of two cases , On Harmonie Ratios in the Spectra of Gases .
341 which is very insignificant , the number of coincidences for the raller limits is in excess .
Table III .
Observed for Calculated from / smaller limit .
larger limit .
0\mdash ; 10 .
48 . . . . . .
43 10\mdash ; 20 .
180 . . . . . .
167 20\mdash ; 30 .
329 . . . . . .
314 30\mdash ; 40 .
478 . . . . . .
444 40\mdash ; 50 . . . . . . . . . .
625 . . . . . .
610 50\mdash ; 60 .
777 . . . . . .
778 60\mdash ; 70 .
886 . . . . . .
882 70\mdash ; 80 .
924 . . . . . .
894 80\mdash ; 90 .
667 . . . . . .
662 90\mdash ; 100 .
253 . . . . . .
263 The fact that the ntlmber of coincidences , though falling short of the calculated values for both limits , is relatively greater for the smaller , suggests the possibility that still narrower limits might give ( results which are still more favourable to the theory of harmonic ratios .
This indeed is the case .
I have counted for all fractions , the of which is smaller than 30 , the number of coincidences for a series of 8limits .
The results are embodied in Table , and show that there is a tendency of the fractions to aggregate into the compartments for smaller limits .
With the exception of the first and last numbers , there is a gradual decrease of coincidences we recede from the harmonic ratios .
Table Limits .
Number of coincidences .
000\mdash ; 095 . . . . . . . . . .
71 095\mdash ; 195 . . . . . . . . . .
85 195\mdash ; 205 . . . . . . . . . .
78 295\mdash ; S95 . . . . . . . . . .
70 395\mdash ; 495 . . . . . . . . . .
68 495\mdash ; 595 . . . . . . . . .
66 595\mdash ; 695 . . . . . . . . . .
56 695\mdash ; 755 . . . . . . . . . .
41 We have now to reconcile two oparently opposite resuIts of our calculations .
On the one hand it was found that the coincidences with harmonic ratios are fewer than we should expect from the theory The following explanation has occurred to me as possibly account .
number ooincidences siven bheory ofnumbers , ence wffect which axistsof poincidences wractions formed berthe formed bigher numbers weduce tumberthe formed bumbers Aoincidence within oiary tbout midway betweenintegers between 1rranged irder ohave taken iccount.nstance , formed bator aumerator ohich aacts.pose tarmonic ratios rexist iumbers , hiefly confined t probability .
If , now , in addition to these coincidences with fractions formed by higher numbers , we uld have a small quantity of real coincidences with the fractions which we have taken into account , the increased quantity of coincidences for small limits over those of larger limits , would be explained .
This explanation might be supported by the fact that , fractiona formed by numbers between 70 and 100 , the coincidences observed more numerous than those calculated on the supposition that the lines are all distributed at random .
It must , however , be remarked that similar eflect might be produced , if any unknown law existed , connecting the lines together , a law which in special cases reduced itseIf to a law of harmonic ratios .
That some law hitherto undiscovered exists I have no doubt , for just in the cases where we have reason to suppose that different lines belong to one system of vibration , we cannot find any coincidencea with harmonic ratios .
The lines of sodium , for instance , are all double ; yet in the set of llines given by Thalen the two componenta approach each other much more rapidly as we pass to more refrangible end of the spectru .
than they would if the Iines were connected together by the harmonio law .
In the additional sets described by Professors Liveing and Dewar no regularity exists in the distance of the two components .
A similar remark applies to the four triplets of magnesium lines .
The triplets resemble each other in so far as the middle line is alwaya nearest to the most refrangible line ; but the resemblance is only a general one , and there is no absolute relation between the distances in each triplet .
On Harmonic Ratios in the Spectra of Taking all these considerations into account , the foIlowing seems to me to be a fair summary of my results for the iron spectrum:\mdash ; .
There is a real cause in a direction opposed to the of fractions formed by numbers smaller than 2 .
eliminahon of the first cause atendency appears ctions formed by two lines to cluster round harmonic 3 .
Most probably some law hitherto undiscovered exists , which in special resolves itself into the law of harmonic ratios .
The subject is of sufficient importance to make further investigation desirable .
We might , for instance , confirm the laws which we have found There anstance , ines ipectrum wmeasurement icases where harmonic ratios heenother spectrum having mhose oanganese ociumBut iseems tromising tncrease tccuracy oaction g\ldquo ; this coincidence to a great degree of accuracy by seeing how far the more gibJe line in the third spectrum coincides with the ] re1 ffingible line in the second spectrum .
Account , of course , must be taken of atmospheric refraction ; reflecting surfaces only to be used .
I hope to try this plan before long , but in order that otheI'S might have the same opportunity , I append a list of all lines which are nearly in the ratio of some fraction formed by integer numbers smaller than ten .
ngstrom 's numbel.s corrected for atmospheric refraction are used .
The table explains itself , but it is perhaps wise to remark again that the number of these coincidences is not larger than one would expect by the theory of probability , and that therefore all of them may prove to be accidental . .
XXX1 .
2 The problem which we have to solve may be stated as follows:\mdash ; Given a certain number of quantities distributed at random betweeu two fixed limits ; form the ratios between every .
of them , and find the expectancy for the number of these ratios which within certain small limits agree with a given fraction .
Iu the first place , we remark that without detriment to the generality of the problem , we may assume the lower of the limits within which all the quantities are lying to be unity ; for if it is not , we may by means of a common multiplier to all quantities reduce it to unity .
Let be the given fraction with which all the ratios are to be compared , and let A be the higher limit which none of the quantities shall exceed .
Assume at first A to be smaller than the square of the reciprocal of .
Divide the range A to 1 into two compartments ; the first from A to , and the second from to 1 .
Let there be quantities which I shall call , . . .
in the first compartment ; and let there be quantities , . . .
in the second compartment .
None of the quantities within one compartment can form amongst them selves ratios which shall be closely coincident with : say ence , if there is only which can range between and 1 , and must lie between and , if there is a coincidence with :he given fraction , the probability of such a coincidence is If there is more than one quantity , , in the first compartment , we observe that these quantities may lie so near together that one and the same can have , within the limits within which we count coinoidences , the required ratio with more than one of the quantities , .
If hese quantities , however , are not sufficiently close together to admit any such double coincidence , the probability that one should have the required ratio with one is .
Call the sum in brackets If we drop the limitation that should not possibly have at the same time the required ratio with more than one , the expression just found will not any more represent the probability of a single coincidence , but it will represent the expectancy for the coincidences .
For in the most general case there is a certain range , which may lie in order to have the required ratio with one of the quantities , ; there a range , , within which a double will happen , and so on : hence the expectancy for the coincidences is but the expression in brackets is always equal to and hence the expression which we have found will represent the expectancy if there is only one quantity , ; for quantities it is We have hitherto supposed that the quantities , , are at given fixed places , or that has a certain given value .
Let , be the probaThe integral represents the expectancy for the sum of quantities equally probable between and , and this expectancy we know be\mdash ; Hence the required expression is\mdash ; In the actual case neither nor are given , we only know their sum ; hence we must add up a number of expressions of the form we have found , varying and , and multiplying each with the proba bility that the particular distribution actually exists .
The that there should be out of values in the first compartmen , is\mdash ; giving successively all values from 1 to we find for the whole expectancy\mdash ; and adding up under the summation sign , the expression reduces to\mdash ; which is the complete expectancy .
We have assumed that A is not larger than the reciprocal of the square of , and we may now extend the formula to larger values of A. Imagine a quantity smaller than and larger than , and Iet A gradually ease from to .
Divide the whole range A into two compartments , one from A to and the second from to 1 , then if a given number of quantities is in each compartment , I can calculate the whole expectancy by knowing:\mdash ; Mr. J. Hopkinson .
Dielectric Capacit.of Liquids .
} 1 .
The expectancy for the coincidences between two quantities in he second compartment .
2 .
The expectancy for the coincidences between one quantity in me first compartment and one in the second .
Now A is supposed to increase gradually from a value smaller than to a value larger than .
As long as it is smaller , the result must pe the same as that we have previously obtained , but none of the uantities which enoer into the calculation show any discontinuity , as a passes through the value , and hence the formula cannot change at that point and must be true as far as the value , or as may be in limit equal to have extended our formula to all values of maller than .
It can be further extended in same way and ust in fact be true for all values of A. .
Dielectric Capacity of Liquids By J. HOPKINSON , F.R.S. Received January 6 , 1881 .
( Abstract .
) These experiments have for object the determination of the refractive indices and the specific inductive capacity of certain liquids , and a comparison of the square of the refractive index for long waves and the specific inductive capacity .
the following table are given the resu.lts obtained for refractive index for long waves deduced by the formula ) , the square { of , and the observed values ( K ) of the specific inductive capacity .
Petroleum spirit ( Field 's ) . . . . . .
Petroleum oil ( Field 's ) . . . . . . . .
( Common ) . . . . . .
Ozokerit lubricating oil Turpentine ( Commercial ) . . . .
Castor oil . . . . . . . . . . . . . . . . . . . .
Sperm oil . . . . . . . . . . . . . . . . . . . .
2 .
Olive oil . . . . . . . . . . . . . . . . . . . .
2 .
Neatsfoot oil . . . . . . . . . . . . . . . .

