The Composition of the X-Rays from Various Metals .
427 small glancing angles , under certain arrangements of the apparatus , an excessive slit width would lead to a fictitiously high ratio of the second to the first order spectrum .
The crystals used in the above investigation were excellent specimens , lent me by Dr. Hutchinson , of Cambridge , to whom my best thanks are due .
I am also indebted to Prof. Brag for the very kind interest he has shown in the work , and to Principal Griffiths , for his sympathy and kindness in considering my requirements .
Further , I desire to express my heartiest thanks to Captain J. H. Shaxby for the valuable guidance and assistance he has rendered me throughout the work .
The Composition of the X-Rays from Various Metals .
By G. W. C. Kaye , M.A. , D.Sc .
, Captain RE .
( T. ) .
( Communicated by Dr. R T. Glazebrook , F.RS .
Received March 13 , 1917 .
) ( From the National Physical .
Laboratory .
) In a paper published in the ' Philosophical Transactions '* in 1908 , the writer described some experiments on the X-rays emitted by a variety of metals when used as anticathodes in an X-ray bulb .
Among the results established was the homogeneity of a large proportion of the X-rays when the bulb was very " soft .
" The absorption curves of the several homogeneous radiations revealed their identities with the characteristic " secondary " radiations which Barkla and Sadler had then recently discovered .
The experiments described in the present paper are an extension of the above , and were carried out partly at the Cavendish Laboratory in 1908 and partly at the National Physical Laboratory just prior to the war .
The writer 's military duties have prevented the continuation of the work , and the results are now put on record in the hope that they may help to further the progress of the subject , to which the more precise method of the crystal-spectrometer has given a great impetus from a different point of view .
Apparatus .
The apparatus employed is sufficiently explained by fig. 1 , which is approximately to scale .
As in the earlier experiments , the various anti* Series A , vol. 209 , p. 123 .
Capt. G. W. C. Kaye .
cathodes were mounted on a trolley , which could be magnetically propelled .
A coil-discharge was employed , the applied voltage being indicated.by an To Electroscope % Screen Screen To Pump Anode Fig. 1 .
alternative spark-gap between polished balls 2*5 cm .
in diameter .
For balls of this size and spark-gaps in the neighbourhood of 1 cm .
, the expression Y = 30,000^ + 2000 gives approximately the voltage Y , corresponding to a spark-length of d cm .
The X-rays passed out of the tube through an aluminium window ( 2 cm .
diameter and 010065 cm .
thick ) , and were received by an ionisation chamber , the front face of which was about 3 cm .
from the window .
The depth of the chamber was 4 cm .
; the sides were of thin aluminium leaf .
A Wilson tilted electroscope was employed to measure the ionisation currents .
The quality of the X-rays was tested by absorption screens inserted between the window and the ionisation chamber .
Metal leaf was used for the thinnest screens .
For details of the various experimental precautions , the writer 's earlier paper on the subject may be consulted .
In the present paper , the radiations dealt with are those from iron , nickel , copper , aluminium , and platinum .
Logarithmic Absorption-curves .
The absorption-coefficient X of a substance is defined by the expression Itf = Io e~xd , where I0 is the initial intensity of an X-ray beam and I\lt ; * the The Composition of the X-Rays from Various Metals .
429 intensity after transmission through a screen of the substance of thickness d. Accordingly , if the thickness of screen ( as abscissa ) is plotted against logio(VIo ) , the slope of the tangent to the curve at any point gives ( when multiplied by 2'3 ) the value of X at that point .
For ordinary heterogeneous X-rays , X diminishes as cl increases , but , if the rays are homogeneous , the graph is a straight line .
Analysis of the Log-absorption Curves .
The method employed in the present experiments was gradually to cut down the X-rays until the log-absorption curve showed that the residual rays were homogeneous or nearly so .
This " end-radiation " was subtracted ( graphically ) from the whole radiation by producing backwards the final straight part of the absorption curve .
A new log-absorption curve was then plotted , employing as ordinate the log of the intensity-difference derived as above .
The resulting graph proved in practice to be either ( a ) a straight line throughout its length , or ( ) a curve merging into a straight line as the thickness of screen increased .
In the latter event , the process of subtraction was applied again .
The underlying notion was to ascertain whether by such analysis the absorption curves could be made to reveal the presence of characteristic radiations .
The justification for the method depends upon its success , for it is possible to express , with very fair accuracy , almost any normal type of absorption curve as the sum of two or three exponentials .
If such components have a physical reality , they will be found to recur in the different absorption curves under varying conditions .
With low discharge voltages on the tube , the dilution of the characteristic rays by general radiation is slight , and the existence of an appreciable amount of a characteristic radiation is usually betrayed before analysis by a perceptible flattening of the absorption curve over part of its course .
But with higher voltages the conditions change , and the method is not so easy unless the accuracy of the measurements and of the graphical analysis is high .
The present experiments are not offered as final , but they do , at any rate , indicate that by such means the detection of component radiations is possible provided they are present in appreciable amounts .
That such homogeneous components may be expected is shown by the fact that the lines in high-frequency spectra are now known to be not nearly so limited in number as was at first believed .
In those cases where the method of graphical analysis is profitable , there is never any difficulty in " spotting " the K and L characteristic radiations ; their absorption-coefficients always come out sufficiently near to the accepted Capt. G. W. C. Kaye .
values .
Incidentally , the method also reveals the proportion of each radiation present in a beam of rays .
In all the log-absorption curves illustrated in this paper , the original curve is indicated by a heavy line , and the resulting components by thin lines .
Most of the radiations were analysed by screens of aluminium , platinum , and copper .
In general , the curves for platinum screens are similar to those for aluminium screens ; those for copper screens do not always share their common characteristics .
Iron Radiation .
Figs. 2 , 3 , 4 show the log-absorption curves of iron radiation in screens of aluminium , platinum , and copper respectively .
In each case the voltage applied to the X-ray tube was 23,000 .
Fe Anh'c^rhode , Al Screen 7mm .
Spa.rk ; \#163 ; 3yOOOvoll\gt ; .
X^-36-4 \X"-258 ( K Ra.dio.rion ) \ 002 003 Thickness of Screen Fio .
2 .
* Aluminium Screens.\#151 ; The absorption curve ( fig. 2 ) is practically straight for screens thicker than 0*02 cm .
There is about 20 per cent , of this homogeneous end-radiation ( X/ pA1 = 13'5 c.g.s. ) , and analysis reveals that the remaining 80 per cent , consists of iron K-radiation ( X/ pAi \#151 ; 95*6 ) .
Platinum Screens.\#151 ; Fig. 3 similarly displays the absorption curve and its resolutes for screens of platinum .
There is about 90 per cent , of Fe K-radiation ( X , / ppt = 360 ) and 10 per cent , of an end-xadiation with X/ pPt = 93 .
of intensify The Composition of the X-Rays from Various Metals .
Fe AnNcaJhode , Pt Screen 7mm .
SpArk 23,000volrs .
Xp\#171 ; 200\lt ; 0 0002 cm .
0 0004 0 0006 Thickness of Screen Fig , 3 .
0-0008 0-0010 Fe Anfic^rhode , Cu Screen 7mm .
Spa.rk ; 23 , OOOvoIte .
0-002 cm .
0-004 0 006 Thickness of Screen 0-008 o-oio Fig. 4 .
Copper Screens.\#151 ; The analysis of the log-absorption curve ( fig. 4 ) again gives about 90 per cent , of Fe K-radiation ( \/ pcn = 100 ) after deducting 10 per cent , of a harder end-radiation with X/ pcu = 45 .
Capt. G. W. C. Kaye .
Figs. 2 , 3 , and 4 are thus very similar to each other and lead to almost identical results .
Nickel Radiation .
Figs. 5 , 6 , 7 are a corresponding set of curves for nickel radiation , again with an applied voltage of 23,000 .
Ni Anticalhode , Al Screen 7mm .
Spa.rk ; 23,000 volt 's .
002 003 Thickness of Screen Fig. 5 .
004 \#163 ; -C o Vte Ni Anficadhode , Pf Screen 7mm .
Sp^rk ; 23,000 volte* \ V mL Ap , \#171 ; 186iT-X/ Ppr-86* \ \| .
* XPf.\#171 ; 58 x/ e* I 0 0002 cm .
00004 0-0006 0-0006 Thickness of Screen Fig. 6 .
o-ooid The Composition of the X-Rays from Various .
433 Ni AnricaJhode , Cu Screen 7mm .
Sp\amp ; .rk ; 2 3,000 volte* -J 05 0004 0*006 Thickness of Screen Fig. 7 .
0-008 0-010 Aluminium Screens.\#151 ; Just as with iron radiation , two components represent the absorption curve ( fig. 5 ) , viz. , 86 per cent , of nickel K-radiation ( \/ pA1 = 58 ) and 14 per cent , of end-radiation with \/ pA1 = 10*8 ) .
This agrees closely with Rawlinson 's figures , * who analysed nickel radiation generated under a voltage of 30,000 , into 79 per cent , of K-radiation and 21 per cent , of rays with \/ pA1 \#151 ; 10*4 .
Platinum Screens.-\#151 ; The absorption curve ( fig. 6 ) is like that for aluminium screens , and we find after subtracting 12 per cent , of end-radiation with Vppt = 86*5 , that the remaining 88 per cent , is entirely Ni K-radiation ( \/ pPt = 270 ) .
Copper Screens.\#151 ; Thus aluminium and platinum screens absorb iron and nickel radiations in similar fashion , but we now find that copper screens treat the two radiations very differently .
With nickel radiation , the absorption curve ( fig. 7 ) is , except for the thinnest screens , straight throughout its length , revealing about 75 per cent , of Ni K-rays .
Not even when the radiation was reduced to per cent , was there any trace of the end-radiation ot figs. 5 and 6 , and we are led to infer that the end-radiation has been abnormally absorbed and transformed into copper K-radiation , the absorption-coefficient of which does not differ very greatly from that of nickel K-radiation .
On the other hand , a soft component 157 ) has ' now made its * 'Phil .
Mag. , ' p. 274 , August , 1914 .
Capt. G. W. C. Kaye .
appearance to the extent of 25 per cent. This may be another characteristic radiation for which A1 and Pt screens are specially absorbent .
It is very probably mixed with the softer copper radiation ( \/ pCu= 100 ) referred to below , which has been excited in the screen by the M rays from the bulb .
Iron radiation is a good deal softer than Ni radiation and the consequent inability of the Fe radiation to excite any such copper component would be an explanation of its absence in the corresponding absorption curve in fig. 4 .
Copper Radiation .
Figs. 8 , 9 , and 10 give the graphs for copper radiation .
The applied voltage in each case was again 23,000 .
Cu Anlicalhode , Al Screen 7mm .
Spo^rk ; 2 3^)00 volts .
00125 cm .
00250 0*0375 Thickness of Screen Fig. 8 .
Aluminium and Platinum Screens.\#151 ; Figs. 8 and 9 show the log-absorption curves of Cu radiation in Al and Pt respectively .
Subtracting 8 per cent , of end-radiation reveals in each case the copper K-radiation as making up the rest of the rays .
The curves are , therefore , precisely similar in type to figs. 2 and 3 for Fe radiation , and 5 and 6 for Ni radiation .
Copper Screens.\#151 ; If now we turn to fig. 10 , we find that copper screens treat nickel and copper radiations very similarly .
Just as with Ni radiation ( fig. 7 ) , but unlike Fe radiation ( fig. 4 ) , the end-radiation completely disappears and is replaced by Cu K-radiation .
At the same time a new and softer component appears with a \jpCxiol about 100 .
The Composition of the X-Rays from Various Metals .
435 Cu Anf ic\amp ; Jhode , Pr Screen 7 mm. Sp\amp ; rk ; 2 3,000 volte \#151 ; 1 0-5 O-0003Cm .
0 0006 00009 Thickness of Screen Fig. 9 .
00015 tig .
11 shows the corresponding curve with a higher voltage ( 45,000)ron the tube .
Both the preceding radiations are again revealed after subtracting a small amount of end-radiation .
Capt. G. W. C. Kaye .
Cu Anric\amp ; rhode Cu Screen l*7cm .
Sp\amp ; rk ; 45,000 volts 0 005 cm .
0 010 0 015 Thickness of Screen Fig. 11 .
0020 The experiment was repeated for a number of applied voltages .
In each case the two radiations were found to be present .
Table I gives the results , and it will be seen that the proportion of K-radiation increases with the voltage while that of the softer radiation diminishes .
It would certainly be of great interest to extend the limits of voltage employed .
According to Whiddington , at voltages below about 10,000 no K-radiation is generated , and it is a matter for enquiry to see if the softer characteristic component here referred to plays a prominent part under such conditions .
The calculated wave-length of this softer component is about 2 x 10~8 cm .
So far as I know , it has not been detected spectroscopically .
The Composition of the X-Rays from Various .
437 Table I. Spark .
Volts .
Radiation .
II 8 A/ PCu \#151 ; 53\gt ; K-radiation .
End-radiation .
mm. per cent. per cent. per cent. 2-5 10,000 32 68 0 7 23,000 26 74 0 10 32,000 24 76 0 17 45,000 16 82 2 p.c. of X/ pCu 7 .
Aluminium Radiation .
Figs. 12 , 13 , 14 give the absorption curves for A1 radiation when passed through Al , Pt , and Cu screens .
The voltage is 23,000 .
Aluminium Screens.\#151 ; The absorption curve ( fig. 12 ) may be analysed approximately into 50 per cent , of end-radiation = 15'5 ) , and 50 per cent , of a radiation having \/ pA1 = 94'3 .
Both values are rather low as there is a small proportion of penetrating radiation present .
For example , the rays which remain after passage through a 1 mm. screen of Al have a ^/ /\gt ; ai = of about 1 .
Capt. G. W. C. Kaye .
A !
Anrica.rbode , Pr Screen 7 mm SfsarK ; 23,000yolrs .
00004- 00006 Thickness of Screen Fig. 13 .
A1 Anricadbocfe , 7com Spark ; 23,000 volte .
0002 0003 Thickness of Screen Fig. 14 .
Platinum Screens.\#151 ; Subtracting end-radiation ( X/ pPt = 95 ) amounting to about 25 per cent , of the total radiation , the remaining 75 per cent , may be represented by a straight line radiation having \/ pPt = 285 .
The Composition of the X-Rays from Various Metals .
439 Copper Screens .
\#151 ; Subtracting the end-radiation ( 15 per cent , in all ) with a X/ pc.i = 26 , the remainder can be represented by a homogeneous radiation with\/ pcu= 168 .
The above results on aluminium radiation are offered only as tentative .
Platinum Radiation .
Figs. 15 , 16 , 17 deal with platinum radiation generated under a voltage of 23,000 .
Aluminium Screens.\#151 ; Fig. 15 betrays little indication of homogeneity until 10 per cent , of end-radiation ( X/ pAi = -5*6 ) is deducted , when a component consisting of 60 per cent , of Pt F-radiation ( A , / pAi = 2o 7 ) clearly reveals itself .
Admixed with this is 30 per cent , of another and softer component which betrays its existence with the thinnest screens .
Further analysis shows that this latter radiation is homogeneous and has a i = TO ( approx. ) .
Pi AnlicbThode , A1 Screen 7mm .
Spd.rk ; 23,000 volte .
_j 05 200(\lt ; \pprox .
) X/ eAr 70 ( approx. ) %Z5'6 XaT64 \VeA7237 ( LR\amp ; .diaJ'lon ) 004- 0 06 Thickness of Screen Fig. 15 .
Platinum Screens.\#151 ; In fig. 16 , the end-radiation has disappeared , and the homogeneity of much of the radiation is apparent .
There is no evidence of the presence of a harder type of ray even though the initial intensity is cut down to nearly \ per cent. The curve is readily resolvable into 60 per cent , of Pt L-radiation ( X/ ppt = 109 ) and 40 per cent , of a softer homogeneous component with X./ pPfc = 330 .
Capt. G. W. C. Kaye .
Pr Anficodhode , Pr Screen 7mm .
Sp\amp ; rk ; 23.000 volts .
El 2340 109 ( L R*di\amp ; J"ion ) 00016 psO OOZO 0 0004 cm 0 0008 O OOIZ Thickness of Screen Fig. 16 .
Pf Anfic'adhode , Cu Screen 7mm .
Spferk ; 23,000 Volfts .
0*0)0 0*008 0-002 Cltl .
0*004 0 006 Thickness of Screen ' Fig. 17 .
Thus both A1 and Pt screens agree in indicating the existence of two homogeneous beams in Pt rays , one the L-radiation , the other with an absorption coefficient three times as large .
The Composition of the X-Rays from Various Metals .
441 Absorption curves obtained at higher and lower voltages , using Pt , screens , disclosed the two components in each case .
The results , which are given in Table II , show that with a low voltage on the tube more than !
half the X-rays consist of the softer component .
An increase in the voltage-leads to an increase in the amount of L-radiation , and a decrease in that of the softer radiation .
Calculation gives a wave-length for the latter of about 2-5 x 10~8 cm .
Table II .
Radiation .
Spark gap .
Volts .
A/ pp^ .
\#151 ; 330 .
Vppt * " L-radiation .
mm. 4 14,000 per cent. 63* per cent. 37 7 23,000 40 60 10 32,000 37 63 * Probably another and even softer component present as well .
Copper Screens.\#151 ; In fig. 17 the absorption curve is built up of 30 percent .
of end-radiation ( X/ /ocu=49 ) and 70 per cent , of Pt L-radiation ( X/ pcu = 140 ) .
We have here an example of the importance of the choice of screen , for whereas a Pt screen allows nothing harder than the L-radiation to pass , , a Cu screen allows nothing softer than the L-radiation to pass .
This behaviour on the part of copper has been noticed by Brag * in his spectroscopic work ; he remarks that the A peak of Pt is completely wiped out by a copper screen .
Ordinary X-Ray Bulbs .
A good many attempts have also been made to resolve the absorption-curves obtained by using A1 screens and ordinary X-ray bulbs ( with platinum anticathodes ) excited by considerable voltages .
The writer 's experience-goes to show that in such cases the end-radiation , even after it is cut down to | per cent , or less , is never really homogeneous .
It may , however , be taken as such for the purposes of approximate analysis .
Such analysis almost always reveals the presence of Pt L-radiation .
In the case of one tube worked at 30 mm. spark , there was 40 per cent , of L-radiation in the rays transmitted through the glass walls of the bulb ; in the case of another at 8o mm. spark , 37 per cent. Most tubes also reveal the presence of a. * ' Roy .
Soc. Proc. , : A , vol. 89 , p. 434 , November , 1913 .
VOL. XCIII.\#151 ; A. Mr. G. W. Walker .
component having A , / pAi = 7*4 , in proportions which increase withi the voltage , at any rate , within a moderate range .
The extent of the work was not competent to throw doubt on the prevailing view that the bulk of the X-rays from a bulb worked under voltages such as obtain in medical practice consists of other than general heterogeneous radiation .
Summary .
The X-rays from a bulb excited by low voltages ( 10,000 to 50,000 volts ) are rich in the characteristic radiation of the anticathode .
In the case of iron , nickel and copper , the amount of K-radiation lies between 80 and 90 per cent. In the case of platinum the proportion of L-radiation is from .
40 to 60 per cent. Evidence of characteristic radiations softer than the K and L-radiation has been obtained .
The writer has perforce had to content himself with little more than setting down observations .
For the preparation of the diagrams he is greatly indebted to Mr. W. E. Higgins , B.Sc. On Magnetic Inertia .
By George W. Walker , M.A. , A.R.C.Sc .
, E.B.S. , formerly Fellow of Trinity College , Cambridge .
( Received March 13 , 1917 .
) In the course of some investigations with regard to the Earth 's magnetism , it occurred to me that a magnetised body must possess magnetic inertia in virtue of its magnetisation , just as an electrified body possesses electric inertia .
I am not aware that this has been considered before , nor have I seen any calculation of the amount to be expected .
The method of dimensions shows that the magnetic inertia of a body of magnetic moment m should be a numerical multiple of C-2 , where C is the velocity of radiation and a a linear dimension of the body the radius , in the case of a sphere ) .
The problem may be investigated and the numerical factor determined by the same general method used for electric inertia.* Consider a conducting sphere of radius a uniformly magnetised with magnetic moment m. * 'Roy .
Soc. Proc./ A , vol. 77 , p. 260 ( 1906 ) , and 'Phil .
Trans./ A , vol. 210 , p. 145 ( 1910 ) .

