]\gt ; .
Nicholson .
Reflect ' on of from [ June 11 is equivalent to about of a cub. mr per day .
Considerin , I a rate of helium production of order of cub. mm. , the agreemen experiment and the theoretical of Rutherford is almost to wonderful , as it does the accuracy of the theory of radio-activ he has done so much to initiate and develop .
I to express my to Mr. Robert Lennox , F.C.S. , an Air .
W. J. Green , B.Sc. , for aid given in conduct of these long an laborious ) eriments .
On the Reflection of from a Stratum of Properties , ?
with to Soum J. W. NICHOLSON , ) , Isaac Newton Student , Scholar of Trinity College , ( Communicated by Professor J. Larmor , kScc . .
Received June ll , \mdash ; Read , 1908 .
In a variable medium , the velocity of ] } ) ation of a train of waves , an the at any are fnnctions of the position of that point .
tances of such ation have only ) worked out in detail ir one particular case .
in connection with the vibrations of a string of variable density , dealt very completely with the in which the density is inversely ) to the distance from fixec point .
fiis original the results applied to optical , and example given .
this rhaps th only case in which a simple exacl solution ) ) edrs possible , ) tion may be to existing , even in , when the ore short ill with the other rned .
The of such a theoly , an { alnination of ) tant cases , is the object of prescnt Lct the velocity 01 a planc -train at some point of a medium , which 1nay choos as that train is 'Theory of Sound , ' ] .
, S 148 . .
Math. Soc vol. ] , 1880 , pp. ; 'Collected Papers , ' vol. 1 , 1908 .
] Stratum of Varying Properties .
The velocity at the point defined by is taken as a function of only .
In order that the results may not be restricted to any particular class of vibrations , let be the vector , of whatever character , whose propagation defines the vibration .
In all } ) ortant cases , may be chosen such a vector that the surface conditions are ( i ) is continuous , ( ii ) is contin uous .
equation of propagation yields for a simple wave-train , ( 1 ) where is the , which is a metion of If is the value of at the origin , we may assume a relation .
( 2 ) When is preponderant over , the approximate solution of the differential equation may be obtained by a method described by H. A. Webb , *after Horn , but employed earlier as a working method by Stokes and L. Lorenz .
Write , ( 3 ) where and are functions of to be suitably chosen .
If accents denote differentiations with respect to , we obtain coefficients of and separately to zero , since the terms are of different orders of nitude , we obtain , if A is an arbitrary constant , and if the first term proves negligible , as will appear in the cases treated , Thus the ooenelal solution , when the variable part of is small with , of the equation , ( 5 ) where represents , is obtained in the form .
( 6 ) We proceed to discuss the reflection backward of waves travelling in traversing a medium of this slowly character , and of extent large compared with the .
Let the medium be unifortn from ' Roy .
Soc. Proc vol. 74 , 1904 , p. ' .
and Phys. Papers , ' vol. 2 , p. 334 .
( Euvres Scientifiques , ' vol. 1 , p. 436 et seq. Dr. Nicholson .
of [ June 11 to , slowly from to , and again uniforn but differently so , from to , there being perfect material con tinuity and , and values of in the rGmG media and A into the atum is of type to which corresponds a refiect c wav from it , The factor , say , is included in K. In the intermediate variable stl.atntl , the complete direct and reflecte disturloance is brriven by ( 6 ) , in final form medium , where there ca be 110 ative wave , .
, If is active index the terminal media in the case , then urface conditions at lead to , .
If , where small pure } inaries of the ordel of a , then Thns , ( 10 which ] lives t amplitude and ) hasc of the eflected wave .
At we which and are real .
( of the wave-length ) small conn ) Rred with that of the distanc this resnlt takes a simpler Write , ( 1 ] .
so that antl is purely real .
lf is the of intensities 01 the reflected incident waves , ; 908 .
] Stratum of Properties .
lso the increase of phase in the reflection , measured at the front , is rhere .
( 13 ) Jigher nntions may be found desired ; the value the function reviously used be found to a order by writing , in the equation propagation , . .
( 14 ) of tl , The result may be tested by coml ) arison with Lord 's solution or a particular previously referred to , which is mathem tically exact .
In that case where is the refractive index between the extreme media , which is to be much different from unity .
Thus I , ( 15 ) while Lord Bayleigh 's result is that the reement is close even for only moderately large values of an Absorbing If the second medium have an absorbent effect , the differential equation relating to it has a term proportional to the velocity , and becomes , ( 16 ) where is the velocity of propagation , and is a certain constant whose value must be positive .
A ative value would denote instability of the If ( has a time factor , and is the -length at the as before , , ( 17 ) and absorption may be accounted for by the function as complex .
But the imaginary tion be essentially negative .
If , therefore , the second medium be ) , and therefore , is complex , and we may write whel e and are positive .
Cf .
a discussion by Lord , of ' Tbeory of Reflection 1 Opaque Bodies , ' ' il .
Mag vol. ] , p. 441 ; 'Collected ks , ' vol. 1 , 1 ) .
14 Dr. Nicholson .
Jieflection of from [ June 11 Thus , in previous notation , cxp Even when is fairly small the function may be very large , the wave-length be small , and ately , in this case , The ratio of the intensities of the leflccled and inciclent waves becomes li .
( 18 A very moderate amount of bsorption is cient , in these to render the reflected wave practically independent of it , and to destro .
entirely the periodic effects with ( thickness of ) , which obtain the absence of value of in ( 18 ) 1nay be called " " " " intensity .
In order to obtain a ) of the validity of this reasoning , it sufficient to consider the case of variation of the medium , which will usually occur in practice .
The ount of variation in the medium , this analysis can take accoun , nrust first be estimated .
In obtaining ( 4 ) , was neglected , then became The ratio borlle to by the greatest term retained in the differentia equation is of order ; and for a law of variation given ) where is distance from the origin , this ) omes , or practically , on reduction , is th " " lneall\ldquo ; if is so reat that ) clror involved is not greatel than about 6 per cent. This may taken as an extrenle case in which the analysis is approximately correct .
If is entirely due to , it ) itten --here is real positive , Thus if is the of ) btratnul , contains a factor -lxp value is ) ) ) in the evtreme case , where is the \ldquo ; leng ) .
This is very even if is a .
The result obtained for is therGfor justified under circumstances .
1908 .
] Stratum of Varyiny Properties .
The formula for the transmitted wave discussed below indicates that in a case like the present the intensity of that wave contains a factor , or when the absorption is extreme .
It therefore appears that the disappearance of the periodic effect with varying thickness of lnyer corresponds , in general , to an absorption is very complete in less than a ength .
The analysis has been shown capable of taking account of this small thickness of provided the rate of change be not too rapid .
For any wave-length there is a amount of change to which the method applies , iven by within an error of 6 per cent. A greater absorption requires a smaller mean wave-length .
Tlbe ted Wave .
In the more general case with previous notation , it may be readily shown that , ( 19 ) which determines the amplitude and phase of the tran mitted wave .
When is real , we may , and also in comparison wibh uIlity .
Thus and the incident wave , , leads to a transmitted undulation , .
( 20 ) When is complex of the form , and is positive , contains a factor , , making very small .
Since this expression in the numerator of ( 19 ) will completely outweigh the smallness of the denominator , there is no appreciable transmitted wave .
It is , in , wholly absorbed , in accordance with the character then possessed by the second medium .
Discussion of the csults .
everting to the case in is real , to which corresponds the formula ( 12 ) , we note that the intensity of leflection , for thickness of the layer , has maxima and minima fter the manner of Newton 's in optics .
But since is essentially positive , olute extinction of the reflected waves is only possible when this quantity is zero , and therefore when The possible laws of slow variation to enable this to happen apart those of periodic character are contained in constant or , ( 21 ) Dr. Nicholson .
Iteflection from [ June 11 , which lcads only to the case diseussed by .
For all other laws , and minima of reffection occur the thickness of the layer varics .
of phase in ( 13 ) is ) sured at the plane , where the waves enter the second mcdium .
If the is unaltered , except by reversal , co whence ) on lting , where is constant and is index of refraction at a point .
This equation caIl be by real values if does not exceed .
) When reflectiou is ttained , to complete but very gradual absorption in the second medium , the of phase must bc .
This is a case in which absorption takes place almost entirely in a Reflection of The results have some ) upon the question of the action of a fog upon incident sound waves .
The ations of Tyndall* and Henryt showed that sounds of dilferent their order of effectiveness , at a distance , in a remarkable manner .
Henry certain conjectures as to the motion of the air , and regarded the consequent refraction as the causing the peculiarities ) distant sounds .
Tyndall , however , whose view seems to be in with many of the obsel.vations on fogsignals , postulated the existence of .
{ Ioccnlent ition of the osphere in such cases , caused by unequal heatill or } the presence of an excess of lnoisture in certain parts .
The views of Stokes and Hellry have been independently by who points out that , since ises sound by it to movcl below than above , any other catlbo such a differencc elocity of must also lift the a is in the tnlosphere .
the results of 's balloon observations has the of tinle by quarter mile he been ) to vscend : feet .
' rnnh .
, p. 18,3 .
' of U.S. trcl for ' lStokes , ' , p. 110 . .
PIOC vol. , 1874 , ] ) .
) ' ' .
1908 .
] of dually VProperties .
Although Peynolds has pointed out that some of Tyndall 's own obseryations ulay be explained by refraction , yet Lord has cited others appear explicable only by acoustic clouds .
, the explanation by refraction of the increase in the of sound when the sky cloudy , or when evening.approaches , appears to of alternative .
An atnlosphere with much aqueous vapour present a greatly power of , as well as of heat , and the removal of the sun 's eHccG therefore produce the acousti phenomenon in question .
The above allow an estimate of the of { iciency of the cause rned by Tyndall .
A portion of the numerator of the expression obtained for the reflected intensity is periodic , and depends both ) the period of the sound and the deglee of heterogeneity of the medium .
For the present we are ecting the radiation of heat takes place dnring the of sound .
changoe in does not appreciably modify the denominator of , but can cause a great alterstion in the numerator : and sounds whose periods are not far apart can thus exhibit , under certain circumstances , a fair amount of difference in etfectiveness .
This is in accordance with observations .
Since , moreover , is large compared with the part of , a fairly small amount of variation of the medium , not greatly and , may cause a more considerable change in which may ffice to change the of effectiveness of two sounds .
A moder small in may cause to oscillate between the limits , so that the corresponding intensity of reflected sound oscillates between the limits , ( 23 ) which may differ fairly widely .
This line of reasoning seems capable of many of the vagaries of sounds coming from a distance .
In weather the atmosphere is more homogeneous than usual , and the backward is extremely snlall .
The sounds then little tendency to vary their ordel effectiv eness , and are also effective at much reater dQtances than ordinalily .
" " acoustic clouds\ldquo ; were farded by him ils mainly dne to the crlCG of of queous vapour in some of the sphele .
, tloist .
has a of heat than dry air , and the qncnt " " stifling\ldquo ; of sound very , although in air under ordinary conditions the effect is ) Tynd ( not give definite idea of the , nature of ctio of acoustic cloud , which may the ) poses of ilis t , by the of Sound , ' 1896 , Stakes , ' Phil. Mag ) , 18 ; Mntl ] .
and ] } , 1 ) .
14 Dr. Nicholson .
Reflection of from [ June 11 , sonlld , by it back to its .
point .
Probably the first ffect both play their ) ) arts .
Moreover , as we have , the vard reflection when so1nd enters such a dissipative nnedium tends to lose its , and to be independent of the reflecting llediumvhen the dissipation exceeds a certain limit .
The of Stokes that the of the ordinary equation of wave tion is to be ) by , where is the period of sound , is the radiation constant .
* Since is snlall pared with , as proved by Stokes , and , it is to be replaced by .
cahG , will Slll ) pose that is zero in air , and that the varial ) mediun of theory is the layer of transition between air and acoustic cloud .
the transition to take place uniformly , we may replace by oneits tinal .
Thus , is to bc lepl by .
Now , the general effect of such complex value of has been already to ] ) tendency towards a certain " " " " reflection of the sound .
The value of the lantity resent , bears to the value , when , the roximate ratio In ( the cllecC of ) that of the variation of is itself snlall , may ttlite .
The ( the exponential therefore btcomc ) V. If this only , there is a reflection is almost limiting , since is ) , a value great the above theory non-periodic reHection to ) .
In the absence of evidencet is for a eloud lpoSed 1 of vapour , it } ) ) ears that acoustic cloud act in lnannel , if the layer of th sulliciently In ( the ) ) wonld seem capable of nation 1 his .
The of Civeness of sounds noticed 011 durin ) ) is change }coustio clo , the on may to found .
Lord ointed o the scuce of an acoustic not inllncncc so of vely tion , as the ) of Cf .
, ' of Sound , ' ol .
2 .
I the 1908 .
] Stratum of dually Varying Properties .
a , in the same manner as the proJonged sound given out by a siren .
Moreover , the diffraction round obstacles may be less effective for sounds of short duration .
These effects of the duration of the sound are , however , independent of the considerations treated at present .
Upward of Sound in the The presence in the ) of a tempcrature gradient , and of a variation of density due to gravity , each cause the circumstances of propagation of sound to vary with the .
When sound travels upward , each will cause a certain fraction to be reflected back to the .
We may consider the effects separately , since both are small , and will first treat that of the temperature gradient .
Let be the velocity at a height , and that at the earth .
If , are the corresp d pressures and densities , then in convective equilibrium constant , and Since we deduce .
( 24 ) If is the period of the sound , and whele Hence , since at the surface , , the intensity ratio of sound reflected from the stratum of height , by the formula ( 12 ) , becomes Since is very small , we .
nay write , retaining only its lowest powers , If is -length at the of the earth , there is extinction of reflected sound when , ( 26 ) where is an integer .
For air under ordinary conditions we write per second , and Dr. Nicholson .
flection of [ June 11 , whence we obtain , ( 27 ) .
and are mcasured in feet .
The total reflection due to the is thus very small , is a eliodic function of amplitude incleas slowly with in a linear manner .
This discussion of case of convecti a general idea of the order of magnitude and mode of variaCion of the effect other distributions of temperature in the of of on lVith the tlsual notation , be ) ) , the equation of } ) ward agation of sound in atmosphore is where is displacement , and , by Boyle 's ] Thus .
( 28 ) ' is the at the SUl.face of the earth , where the velocity is , then or , ( 29 ) where .
( 30 ) The of the formula ( 6 ) is alJd the is or .
Moleover , .
is the ibility .
Thus ( refole 1 the formula ( 12 ) the intensity of sound reflected from is .
( 31 ) undcl conditions , ) his small .
1908 .
] Stratum of radually Properties .
the calculation , it appears that for sound whose wave-length , expressed in feet , is .
( 32 ) This effect is of the same character as that of the temperature gradient .
For a given elfect of the variation of density is about 700 times as great as that of the temperature variation , provided that the ratio of height to wave-length is great .
The type of analysis here employed is applicable to many other pbysical problems connected with short waVes , such , for examl ) , as that of short waves of transverse displacement in an infinite stretched string .
TDIx.\mdash ; On Viscosity , onduction , and their Bffect on the Propagation of in a This appendix is an amplification of the previous discussion of .
by moist .
It was there tacitly assumed that sipation of sound took place merely by radiation of heat during its .
This was based on the fact that radiation is the factor most liable to cause appreciable dissipation in sounds of any pitch whatever .
Other factors capable of causing it for a smaller range of pitch are viscosity and conduction .
If is the kinematic viscosity , and the thermometric conductivity , then the equation of ation of sound , when these dissipative encies alone act , , where .
In C.G.S. units and thus cm .
per second , and for a sound of frequency cm .
It was shown above that the quantity of reflected sound in a val.iable medium , when radiation is present , bears to that when adiation is absent the ratio , or .
Here is the quantity in the dissipation ralio due to radiation lone , which occurs in ordinary propagation in a uniform Kirchhoff , 'Pogg .
Ann vol. 134 , p. 177 , 1868 ; Lord Ra leigh , ' of } edition , S , 'Wied .
vol. 60 , p. 113 , 1896 .
298 Iieflection of ) from Stratum of roperties .
Thns cm .
and Thus if ( 1 bc not less than about 1/ 500 , radiation is the predominant factor in dissipation of sounds of low pitch , and , therefore , also in reflection the mediunl is not uniform .
sound of very pitch , radiation is not so necessarily predominant .
In the case of a soulld of frequency ( approx. ) .
As to the value of , much ence of opinion exists .
Wilmer Duff , * in a series of iments on the late of fall of intensity of sounds of frequency bout 7 , concluded that the great rate obtained could only be explained by to a value in ordinary air .
parently valid reasons were for that atmospheric refraction and interllal reflection had not any ) eciable part in the rate of fall , for the conditions of the atr/ losphere and the circumstances of position appear to have widely .
Morcover , wind was almost and the vertical tetnperature gradient , the other chief cause of refraction , must have been subject to variation , as obscrvations were taken at noon and after sunset .
But Lord Rayleigh , in examinin , these results , concluded that the radiating power must be hundreds of times aller , and attributed Duff 's results to atmospheric refraction , or to some iCherCo nected cause .
He suggested , as a possible cause , a delay in the equalisation of difTerent kinds of energy in a compressed not insensible in rison with a period of sound , and thus causi1lg dissipation .
This cliticism was supI ) ted by an ument based on the theory of , and ) iment on the lines of that used by Clement Desormes in .
The agreement of this experiment with to be and we must therefore ) is not of order unity for dinary air .
This justifies the assumption made in a previous portion of this paper that ( ordiuaxy .
very lloist the value of does not to been mined entally , and hnvG no irect of il , except ) it gteater than in ordinary .
That its valuc ) ) in with that for dry air seems fnirly certain .
) eover , ) viscosity ) conductivity may be , nnd probably are , reater i lloist , thus lending support to the ument on ( 18 ) , adding their effects to those of ( which latter probably ntes , except for sounds of very pitch ) .
' Review , ' March , 'Phil .
Mag vol. 47 , p. 308 , 1899 .
'Phil .
' vol. 4314 .
1899 .
On the Nature of the in the Electric Spark .
299 Until the values of these quantities , and more especially of , are known for very moist air , the explanation iven 1 ) Tyndall of his own observations must be regarded as a very possible one , for the change of character of the reflection effect of ( 12 ) may readily occur in a sufficiently layer , if the three coefficients be at all appreciably modified by the ence of moisture .
We may note that the effect of alone is independent of the pitch .
On the of the in Electric Spark .
By S. B. MILNEB , D.Sc .
( Lond. ) , Lecturer in Physics , the University of Qheflield .
( Communicated by Professor W. M. Hicks , F.R.S. Received February 10 , \mdash ; Read March 5 , 1908 .
) ( Abstract .
) The main subject of the wolk described in the present paper consists in the examination of the streamers in the inductive spark in the matic of the various metallic lines , It thus forms an extension of the research of Messrs. Schuster and Hemsalec which the examination of the streamers was restl.icted to the inductionless spark .
The ) vations were taken by raphing the spectrum as } out by a rotating mirror , the slit of the spectroscope removed and the spark itself , so that each line of the spectrum a monochromatic image of the spark .
In order to avoid the superposition of series of streamers which are for1ned in the out of each nlonochromntic , an arrangement of the prisms of the spectroscope used by which , while the images of the spark on the screen were vertical , out in a horizontal direction , the dispersion of the spectrunn was in a dilection to the horizontal .
this arrangelnent the series of streamers corresponding to each metallic line becomes distinctly separated from the others .
Photographs of the streamers in the spectra of the sparks fionl following metallic poles were taken , in each case with a number of diHelent inductances in series with the spark : aluIlIinium , antimony , bisnruth , cadmium , calcium , copper , lead , magnesium , mercury , nickel , } ) latinunt , 'Phil .
Trans , vol. 193 , p. 189 ( 1900 ) .

