Hydrodyncimical Problems Suggested by Pitot 's .
503 discharge .
Thus No. 1 was obtained with an uncondensed discharge and No. 5 with a heavy condensed discharge .
No. 2 , with a moderate condensed discharge , shows lines of the ordinary carbon spark spectrum with XX , 4650 and 5696 .
In No. 3 , the X 5813 group and X 5593 are just visible , whilst No. 5 shows these lines with considerable intensity ; they are situated in a part of the spectrum to which the plates used are comparatively insensitive , and their photographic intensities are , in consequence , very small in comparison with their appearance in visual observations .
In the more refrangible region , which in the plate was necessarily over exposed , strong lines due to oxygen , etc. , from the walls of the tube are visible .
The use of a filter to counteract the sensibility curve of the plate was only partly successful .
Hydrodynamiced Problems Suggested by Pitot 's Tubes .
By Lord Rayleigh , O.M. , F.R.S. ( Received June 5 , 1915 .
) The general use of Pitot 's tubes for measuring the velocity of streams suggests hydrodynamical problems .
It can hardly be said that these are of practical importance , since the action to be observed depends simply upon Bernoulli 's law .
In the interior of a long tube of any section , closed at the further end and facing the stream , the pressure must be that due to the velocity ( v ) of the stream , i.e. \pv2 , p being the density .
At least , this must be the case if viscosity can be neglected .
I am not aware that the influence of viscosity here has been detected , and it does not seem likely that it can be sensible under ordinary conditions .
It would enter in the combination where v is the kinematic viscosity and l represents the linear dimension of the tube .
Experiments directed to show it would therefore be made with small tubes and low velocities .
In practice a tube of circular section is employed .
But , even when viscosity is ignored , the problem of determining the motion in the neighbourhood of a circular tube is beyond our powers .
In what follows , not only is the fluid supposed frictionless , but the circular tube is replaced by its two-dimensional analogue , i.e. the channel between parallel plane walls .
Under this head two problems naturally present themselves .
The first problem proposed for consideration may be defined to be the flow Lord Rayleigh .
of electricity in two dimensions , when the uniformity is disturbed by the presence of a channel whose infinitely thin non-conducting walls are parallel to the flow .
By themselves these walls , whether finite or infinite , would cause no disturbance , but the channel , though open at the fihite end , ' is supposed to be closed at an infinite distance away , so that , on the whole , there is no stream through it .
If we suppose the flow to be of liquid instead of electricity , the arrangement may be regarded as an idealised Pitot 's tube , although we know that , in consequence of the sharp edges , the electrical law would be widely departed from .
In the recesses of the tube there is no motion , and the pressure developed is simply that due to the velocity of the stream .
The problem itself may be treated as a modification of that of Helmholtz , * where flow is imagined to take place within the channel and to come to evanescence outside at a distance from the mouth .
If in the usual notation^ z == x-f iy , and w = \lt ; f\gt ; + iyfr be the complex potential , the solution of Helmholtz 's problem is expressed by Z ( 1 ) or x \#151 ; \lt ; f\gt ; \#151 ; 6^cos yj/ " , ( 2 ) The walls correspond to yfr = +7r , where y takes the same values , and they extend from x \#151 ; \#151 ; 00 to a ; = \#151 ; 1 .
Also the stream-line ^ = 0 makes = 0 , which is a line of symmetry .
In the recesses of the channel \lt ; f\gt ; is negative and large , and the motion becomes a uniform stream .
To annul the internal stream we must superpose upon this motion , expressed say by \lt ; j\gt ; 1 + iyjn , another of the form \lt ; j\gt ; 2 + ^2 where ( f\gt ; 2 -f \#151 ; \#151 ; x\#151 ; iy .
In the resultant motion , \lt ; f\gt ; = \lt ; f\gt ; \ + \lt ; f\gt ; 2 = \lt ; f\gt ; 1\#151 ; yfr = yjri~\-yjr2 = ^1\#151 ; yr \#166 ; so that \lt ; f\gt ; \ \#151 ; \lt ; f)-\-x , \#151 ; yfr-by , and we get 0 = \lt ; \#163 ; -j-e*+x cos 0 = sin ( 3 ) whence # = \#151 ; \lt ; \#163 ; + log \/ ( ^\gt ; 2 + Vr2 ) , \#166 ; ^ + tan"1(i|r/ \lt ; ^ ) ( 4 ) or , as it may also be written , z \#151 ; \#151 ; w log w. ( 5 ) It is easy to verify that these expressions , no matter how arrived at , satisfy * ' Berlin Monatsber .
, ' 1868 ; ' Phil. Mag./ vol. 36 , p. 337 ( 1868 ) .
In this paper a new path was opened .
t See Lalhb 's 'Hydrodynamics/ S66 .
Hydrodynamical Problems Suggested by Pitot*s Tubes .
505 the necessary conditions .
Since x is an even function of yfr , and y an odd function , the line y \#151 ; 0 is an axis of symmetry .
When = 0 , we see from ( 3 ) that sin = 0 , so that y\#151 ; 0 or +ir , and that cos^ and \lt ; f\gt ; have opposite signs .
Thus when \lt ; f\gt ; is negative , y\#151 ; 0 ; and when ( f\gt ; is positive , + 7r .
Again , when ( f\gt ; is negative , x ranges from + oo to oo ; and when \lt ; f\gt ; is positive x ranges from \#151 ; oo to \#151 ; 1 , the extreme valde at the limit of the wall , as appears from the equation dx/ d(f ) = \#151 ; 1 + 1 = 0 , making 0 = 1 , x = \#151 ; 1 .
The central stream-line may thus be considered to pass along y \#151 ; 0 from x \#151 ; co to X \#151 ; \#151 ; oo .
At x\#151 ; \#151 ; co it divides into two branches along y \#151 ; + tt .
From x = \#151 ; oo to \#151 ; 1 , the flow is along the inner side of the walls , and from x= \#151 ; 1 to x= \#151 ; oo back again along the outer side .
At the turn the velocity is of course infinite .
We see from ( 4 ) that when yfr is given the difference in the final values of y , corresponding to infinite positive and negative values of \lt ; jE\gt ; , amounts to it , and that the smaller is yjr the more rapid is the change in y. The corresponding values of x and y for various values of \lt ; f\gt ; , and for the stream-lines yjr= \#151 ; 1 , \#151 ; are given in Table I , and the more important parts are exhibited in the accompanying plots ( fig. 1 ) .
Table I. \lt ; p. 4 = * = -4- ^ = -1 .
X. * X. y. X. y- -10 12-303 0 -2750 12-30 0-550 12 -31 1 -100 - 5 6 -610 0-3000 6 -614 0-600 6-63 i 1 -198 - 3 4 -102 1 0 -3333 4-112 0-665 415 1 -322 - 2 2-701 0 -3745 2 -723 0-745 2*80 1 -464 - 1 1 *030 0-495 1 111 0-964 1 -35 1-785 - 0 50 ' 0 -081 0*714 0 153 1-285 \#151 ; - 0-25 -0-790 1 -035 \#151 ; \#151 ; \#151 ; .0-00 -1 386 1 *821 -0-693 2 071 o-oo 2-571 0-25 -1 -290 2 -606 \#151 ; \#151 ; \#151 ; \#151 ; 0 '50 -1-081 2 -928 -0-847 2 -881 - 0-388 3 035 1 *0 -0-970 3 -147 -0-888 3-178 !
\#151 ; 0653 3 -356 2 0 -1 -299 3 -267 -1 -277 3-397 j \#151 ; 1 -195 3 678 3 0 -1 -898 3-308 -1 -888 3-477 \#151 ; \#151 ; 4*0 \#151 ; \#151 ; \#151 ; \#151 ; - 2-584 3-897 5 0 -3 -389 j 3 -342 1 -3-386 !
3-542 \#151 ; \#151 ; 10*0 -7-697 3-367 ; _ - 7 -692 4 -042 20 0 ; \#151 ; \#151 ; ; \#151 ; -17-00 4 -092 In the second form of the problem we suppose , after Helmholtz and Kirchhoff , that the infinite velocity at the edge , encountered when the fluid adheres to the wall , is obviated by the formation of a surface of discontinuity , Lord Rayleigh .
Fig. 1 .
Hydrodynamical Problems Suggested by Pitot 's Tubes .
507 where the condition to be satisfied is that of constant pressure and velocity It is , in fact , a particular case of one treated many years ago by Prof. Love , entitled " Liquid flowing against a disc with an elevated rim , " when the height of the rim is made infinite.* I am indebted to Prof. Love for the form into which the solution then degrades .
The origin O ' ( fig. 2 ) of or z is taken at one edge .
The central stream-line ( yjr == 0 ) follows the line of symmetry AB from y = H-oo to y \#151 ; \#151 ; co .
At y = \#151 ; oo it divides , one half following the inner side of the wall CO ' from \#151 ; oo to = 0 , then becomes a free surface O'D from y \#151 ; 0 to y = \#151 ; oo .
The connection between z and w ( = \lt ; f\gt ; 4- tie ) is expressed with the aid of an auxiliary variable 0 .
Thus z \#151 ; tan 0\#151 ; 0\#151 ; tan2 0\#151 ; log cos 0 , ( 6 ) w = \#163 ; sec2 0 .
( 7 ) If we put tan 0 \#151 ; t ' + iy , we get w = i(l + r\gt ; -V+2 ifr so that \lt ; \#163 ; = |(1 + p-772 ) , == \#163 ; ( 8 ) We find further ( Love ) , S = \#163 ; + iy + \#163 ; \#163 ; y\#151 ; 2(\#163 ; 2\#151 ; y2)\#151 ; btan-1-\#151 ; \#151 ; \#151 ; \#163 ; tan"1 \#151 ; 2 4 1 1 + f2\#151 ; V + ilog{(l-^ + f } ( 9 ) sothat ^ = f+'f+ |tan~1 2f +|taa~ ' 4--^ , ( 10 ) y = ''7 \#151 ; HI2\#151 ; { ( 1\#151 ; \#166 ; \#187 ; 7)2+P } ( II ) The stream-lines , corresponding to a constant yfr , may be plotted from ( 10 ) , ( 11 ) , if we substitute 2yfr/ \#163 ; for y and regard \#163 ; as the variable parameter .
Since by ( 8 ) \lt ; *\gt ; = \#177 ; ( i + F)-^7\#163 ; 2 , = \#163 ; \#163 ; +2*2/ f , there is no occasion to consider negative values of and \lt ; f\gt ; and \#163 ; vary always in the same direction .
As regards the fractions under the sign of tan-1 , we see that both vanish when | = 0 , and also when \#163 ; = oo .
The former , viz. , 2\#163 ; -j- ( 4\gt ; /r2/ |2 + \#163 ; 2\#151 ; 1 ) , at first + when \#163 ; is very small , rises to oo when ^2 = ^{l\#177 ; v/ ( l \#151 ; 16 t/ r2 ) } , .which happens when but not otherwise .
In the latter case the fraction is always positive .
When the fraction passes through oo , there changing sign .
The numerically least negative value is reached when f2 = h{ v ' ( 1 + 48-^r2 ) \#151 ; 1 } .
The fraction then retraces its entire course , until * 'Camb .
Phil. Proc. , ' vol. 7 , p. 185 ( 1891 ) .
Lord Rayleigh .
it becomes zero again when g = oo .
On the other hand the second fraction , at first positive , rises to infinity in all cases when \#163 ; 2 \#151 ; \ 16-\lt ; /r2)\#151 ; 1 } , after which it becomes negative and decreases numerically to zero , no part of its course being retraced .
As regards the ambiguities in the resulting angles , it will suffice to suppose both angles to start from zero with f. This choice amounts to taking the origin of a ?
; at O , instead of O ' .
When yfr is very small the march of the functions is peculiar .
The first fraction becomes infinite when \#163 ; 2 = 4 yjr2that is when \#163 ; is still small .
The turn occurs when \#163 ; 2 = 12 ^2 , and the corresponding least negative value is also small .
The first tan-1 thus passes from 0 to while \#163 ; is still small .
The second fraction also becomes infinite when \#163 ; 2 = 4^2 , there changing sign , and again approaches zero while \#163 ; is of the same order of magnitude .
The second tan-1 thus passes from 0 to 7r , thereby completing its course , while \#163 ; is still small .
When yjr = 0 absolutely , either \#163 ; or rj , or both , must vanish , but we must still have regard to the relative values of yfr and \#163 ; .
Thus when \#163 ; is small enough , x \#151 ; 0 , and this part of the stream-line coincides with the axis of symmetry .
But while \#163 ; is still small , x changes from 0 to 7r , the new value representing the inner face of the wall .
The transition occurs when \#163 ; = 2 yfr , 7 ] = 1 , making in ( 11 ) y = \#151 ; oo .
The point O ' at the edge of the wall ( x = 7r , y = 0 ) corresponds to \#163 ; = 0 , rj = 0 .
For the free part of the stream-line we may put 0 , so that x\#151 ; ftan"x = \#163 ; \#151 ; tan-1 \#163 ; + tt , where tan-1 \#163 ; is to be taken between 0 and Also y = -H2+iiog(i+r\gt ; .
When \#163 ; is very great , x = f+ivr , y = \#151 ; if3 , and the curve approximates to a parabola .
When f is small , X\#151 ; IT \#151 ; ||3 , y \#151 ; so that the ratio ( x\#151 ; 7 r)Jystarts from zero , as was to be expected .
The upward movement of y is of but short duration .
It may be observed that , while dx/ di\ is always positive , a6\gt ; tff 2(i + f2 ) ' , r . .
.v which is positive only so long as f\lt ; l. And when f \#151 ; 1 , \#151 ; 7r = 1 \#151 ; ^7r = 0'2146 , y = \#151 ; ^ + log 2 = 0*097 .
( 12 ) ( 13 ) ( 14 ) ( 15 ) Hydrodynamical Problems Suggested by Pitot 's Tubes .
509 Some values of x and y calculated from ( 12 ) , ( 13 ) are given in Table II , and the corresponding curve is shown in fig. 3 .
Table II.\#151 ; yfr = 0 .
X. y- X. \#166 ; * 0*0 3 142 0 | 2*5 4*451 - 0 *571 0*5 3*178 + 0*050 3*0 4 *892 - 1*098 1 *0 3*356 + 0*097 4*0 5*816 - 2*583 1*5 3*659 + 0*027 5 *0 6 *768 \#151 ; 4*62 2*0 | .
4*034 -0 *195 20*0 21 *621 -97 00 Fig. 2 .
It is easy to verify that the velocity is constant along the curve defined by ( 12 ) , ( 13 ) .
We have dx _ eat die _ .
d\lt ; j\gt ; " l + \#163 ; *\lt ; tie * d\lt ; f\gt ; 21 and when = 0 , = 1 + F ) , d\lt ; f\gt ; / dg = | f. Thus dx 2 f die 1 \#151 ; f2 d(f\gt ; 1 + f2 ' d\lt ; f\gt ; ~ 1-ff2 ' and ( dxjd\lt ; t\gt ; Y + ( \lt ; die / d\lt ; f \#151 ; 1 .
The square root of the expression on the left of ( 17 ) represents the reciprocal of the resultant velocity .
510 Hydrodynamical Problems Suggested by Pitot 's Tubes .
Table III.\#151 ; = \#163 ; .
J .
| y- S. X. y , r o !
o 00 0-40 | 2 -9667 + 0 076 0 05 o 1667 9-098 0-50 3 0467 0-130 0 10 0 *2995 3-008 0-60 3 1089 0-162 0-13 !
0-4668 1 -535 0-80 3 -2239 0 -198 0-15 !
0 -6725 0-766 1 -oo 3-3454 0 -207 0-17 i 1 0368 + 0 -109 1 -50 3 '6947 + 0 -125 0-18 j 1 -2977 -0 143 2-00 4-0936 -0*112 0-19 !
1-5907 -0-304 2-50 4 -5234 -0*501 0-20 !
1 -8708 -0-370 300 4 -9725 -1 *032 0 *22 i 2 -2828 -0 -331 4-00 5-9039 -2-536 0-25 1 2 -5954 -0*195 6-00 7-8305 -7 -161 0-30 1 2 -8036 -0-047 1 Fig. 3 .
Fluorescence and Resonance of Sodium .
511 When ylr differs from zero , the calculations are naturally more complicated .
The most interesting and instructive cases occur when yjr is small .
I have chosen \fr = 1/ 10 .
The corresponding values of f , and y are given in Table III , calculated from equations ( 10 ) , ( 11 ) , and a plot is shown in fig. 3 .
As in the former problem , where the liquid is supposed to adhere to the walls notwithstanding the sharp edges , the pressure in the recesses of the tube is simply that due to the velocity at a distance .
At other places the pressure can be deduced from the stream-function in the usual way .
Observations on the Fluorescence and Resonance of Sodium Vapour.\#151 ; II .
By the Hon. R. J. Strutt , Sc. D. , F.R.S. , Professor of Physics , Imperial College , South Kensington .
( Received June 12 , 1915 .
) [ Plate 8 .
] S 1 .
Introduction , Prof. R. W. Wood has made many interesting observations on the fluorescence of sodium vapour .
They are conveniently summarised in his ' Physical Optics ' ( MacMillan , 1911 ) .
The fluorescent spectra he has obtained are apparently connected with the banded absorption spectrum of dense sodium vapour .
If white light is employed , this banded absorption spectrum is re-emitted completely as a fluorescent emission spectrum .
If monochromatic light is used , a portion only of the complete band spectrum is emitted , thik portion consisting of a moderate number of lines approximately equally spaced along a normal spectrum , and including a line coincident with the exciting line .
These observations on the band spectrum are of great interest and importance , but observations on the line spectrum of sodium , in absorption and in fluorescence , are more within the range of theoretical discussion at the present time .
The line spectrum is observed in absorption and in fluorescence at a density of sodium vapour small compared with that needed for the band spectrum .
If a beam of white light traverses such vapour , the D line and the other lines of the same series in the ultra-violet are seen in absorption .
At the

