]\gt ; Prof. J. C. McLennan and Mr. R. J. Lang .
The various parts with their designatin letters are as follows:\mdash ; A. Tinned brass tube .
B. Ground end plate .
C. Grating carriage .
I ) .
Rails for carriage .
E. Ruled diffraction grating ; 20,000lines per inch , 120 cm .
radius .
F. Tube with ground fitting for exhaust .
G. Large round tap containing film carrier .
H. Photographic , Schumann sensitised .
I. eye-piece .
J. Glass right-angled prism .
K. L. Exhaust tube for film carrier .
M. Aperture for loading film carrier .
N. Ground fitting for eye-piece , replaced by ground plug in working .
Ground taps for exposing through slit .
Q. Screw for adjusting vertical inclination of grating .
R. Micrometer screw for setting orientation of grating . .
Mount for diffraction grating .
At the slit and film end of the tube on one side there was a mount for an .
eye-piece ( I ) and prism ( J ) mounted in a taper plug ( N ) .
When not in use , the plug ( N ) with eye-piece and prism were withdrawn complete and replaced by a solid plug .
An Investigation of Extreme Spectra .
263 In practice it was found impossible to keep the ground joint made by the tap with its matrix perfectly airtight at evel y point without all exposed parts of the joint with wax .
In the tap to expose a plate , this wax coating was always broken , wirh the result that leaks occurred .
After spending considerable time trying to rectify this defect , the best results were finally obtained by abandoning the idea of exhausting the plate-holder space in the tap preliminary t , o turning the latter so as to expose the ate .
The tap as originally provided with the apparatus , was replaced by a second one , shown in fig. 4 .
This large tap carried the plate-holder as the original one did , and instead of the plate-holder being inserted in a recess in the side of the tap , as formerly , the tap was slotted out , to one end , as shown at " " \ldquo ; in .
This the plate-holder , mounted on a thin brass strip , to be inserted from the end A When the plate was inserted , the was covered with a small brass disc , which was sealed about its edge with wax .
With this arrangement the tap always had its plate-holding recess facing the grating and ready for an exposure .
The exposed parts of the tap joint were permanently covered with a wax coating , to secure airtightness .
FIG. 4 .
The use of the special tap necessitated the of the whole apparatus each time an exposure was made .
The Trimount pumps , however , were very rapid and efficient action , and it was found comparatively easy to reach with half an hour 's continuous running a pressure low enough for the induction coil discharge to give green fluorescence in the pressure-testing vacuum tube .
III .
of C'oncave Grating .
With a concave rating all spectra are focussed upon a whose diameter is the radius of curvature of the grating .
As stated above , the of the spectrograph was bent to the arc of such a circle .
In fig. 5 let SPC be a section of the plate ; a section of the rating , with the normal CA ; AB the distance between two lines on the grating , or the grating constant ; S.the source of light , and a line of -length , and of the first order .
Prof. J. C. McLennan and Mr. R. J. Lang .
Since the total length of the plate was only 8 cm .
it is clear that would lie between and C. SA , PA Total difference of path is .
Now let \ldquo ; be the distance of the line " " \ldquo ; from " " \ldquo ; measured along the plate and let , ' \ldquo ; be the distance from grating to the plate .
or Therefore last term we get Now is very nearly equal to unity since " " \ldquo ; is always small compared with " " \ldquo ; and also may be taken as Then .
If now be taken as constant , i.e. , if the grating is kept fixed , we have straight line curve between and " " Its slope is If the be turned to bring an unknown region upon the plate , the of the calibration curve is altered , so that the curve is not a continuous straight line .
We have therefore , a new slope to the curve every time the rating is turned .
Now if after the grating is turned lines of known wave-length appear on the plate , then we have , , where above .
of Now slope of the line was shown to be go that one does not need to know anything more than distance between the two lines and the values of their wave-lengths to plot a calibration for the plate for a criven setting .
IV .
Nethod I re-le The method used to find the wave-length of the lines observed was as follows:\mdash ; The zinc spark line .
was obtained on a plate , and its distance from the short wave-length end of the plate ( which always occupied the same position in its holder ) measured .
A mercury arc spectrum plate was then taken and the distance of the line A.U. found from the end of the plate .
Thus the distance between the lines .
and .
was found .
A large sheet of graph : having one space for each A.U. and one each mm. dispersion was procured .
The line A.U. o give zero dispersion , and the line .
a second point .
A straight line ) dlaw1l through , these two points .
This raph was used to obtain of all lines obtained this setting of the grating .
The two carbon lines .
and 4 vere measured in this way .
Then the .
was turned and another carbon plate taken which contained these two hnes others .
The line .
was riven zero dispersion , and the dispersion ot .
measnred on this plate .
A new line graph If was obtained having a htly different slope from the last .
the grating was turned , and this the lines .
and were used to get the new graph .
Finally , for the last change of the the lines .
and A.u. were used .
This method , it is ' realised , had defects , but it is probable that it enabled us to measure wave-lengths within one ngstrom unit .
of Work , fi'xtreme Ult , -riolct The first successful measurements of wave-lengths in the Schumaltn were those so carried out by Prof. with a concave retiection grating at Harvard .
grating sectl'osco used consisted roughly of the spectrosco } ) itself , and an case or container .
The grating was fixed in one end of the brass tube constituting * Lyman , " " Preliminary Measurement of : ' Nature , ' vol. 69 , p. 465 ( March 17 , 1904 ) .
V0L .
7 .
Prof. J. C. bIcLennan and Mr. R. J. Lang .
spectroscope upon an djnstable mounting .
In the other end was a draw-tube carrying the slits and plate holder .
The latter was designed to take a number of exposures successively by means of a novel magnetic arrangement which shifted the plate while the vacuum was still maintained .
The , had a radius of 97 cm .
, and 15,028lines to the inch .
It was ruled with the object of throwing as much energy as possible into one order .
The spectra chiefly studied were those of hydrogen and helium , and the sources were end-on dischal.ge tubes .
The tubes verc separated from the spectroscope by a fluorite window until the limit of the fluorite was reached when the window was discarded and placed in both the tube and the cope .
The spark spectra of some metals were investigated by causing a spark to pass between electrodes of the metal in an atmosphere of These spectra were not carried beyond the limit of the fluorite window because it was found ense atmosphere inthat arare aphere lpectroscope wydrogen gothing bydrogen lines , while parent to the radiations .
The possibilities of the spectrum of hydrogen given with the tube reached a limit near 900 and helium was It is fortunate that both these ases , which according to the Bohr theory have a series of lines in the Schumann , should be transparent to extremely short waves .
Prof. Lyman found more than 300 lines in hydl O , the test about 900 .
The first three lines of the series predicted by as an analogue to Balmer sel'ies , namely , , and rdinary disch eared , hortest beady found iogen , strong d .
This appears to confirm the Bohr theory , that the helium atoln , : lost but one electron , should upon turll show a spectrum very similar to hydrogen , but that when both ] ectronb are returning it should a spectrum of its own .
; 4 Besides this work in the arc spectra of , carbon monoxide , S , calcium , Croutitlm , and lnercury were extended to abotlt while the spar pectra of mercury and aluminium were : carried to Besides the work of .
Lyman , that of Handke* and ) mentioned .
The former worked with spark and ucceeded in zinc , copper , tin .
silver , and fold to , approximately the latter carried the arc spectra of mercury , cadmiu1n , and zinc to ) .
The opacity of , of course , linlited both ) experiments .
In a recent note on ' The Limit of the Spectrum in the Prof. Lyman describes some later experiments .
In these } } used a grating of 50 cm .
radius in place of the 100 cm .
one previously mployed .
The of the quartz discharge tube was altered so that the end of tl capillary could be brought near to the slit .
potential difference of the transformer was also increased .
Using heliunl mm. pressure .
extended the spectrum of the from 600 .
to .
A trace of line was also found on his plates at 450 , but it was far too faint , in view , to afford trustworthy evidence .
He also used Wood 's miniatnre arc vacuum , and a variety of vacuum spark ements , but none of sources appeared to yield lines in the most refrangible region .
VI .
Use ) The work done in the present investigation has Leen confined .
to alcs of carbon , mercury , iron , copper , and zinc .
It was found that an arc of a metal having a high fusion point may be maintained in a vacunlm i means is employed to start it by bringing the electrodes together then them .
The carbon arc was found to be the easiest to maintain and also to give the OIlgest radiations of very short wave-lengths , so that its spectrum has been oarried the farthest into the unknown region .
The first lamp used was merely a circular glass bulb about 6 inches in diameter , attached to a brass tube with Kotinsky cement .
The tube was attached to the slit of the pectroscope .
The carbon electrodes } introduced through the glass tubes in the sides of the bulb , one directly towards the slit and the other at right angles to the first .
A ) iece of rubber tubing over the electrode tubin rendered the joint airtight , and still allowed enough freedom to make and break the arc .
lamp was not a success , because the intense heat destroyed the glass .
AIso , since the acuation of the lamp took place through the slit , ticles from the vapours became deposited on the sides of the latter and tended to truct the passage of the light .
This latter difficulty was overcome b ) attaching permanently ovel the slit a short tube to which the laml } * Handke , ' Inaugural DisseItation , ' Berlin , 1909 .
Lyman , ( Science , ) vol. 45 , p. 187 .
2 .
this an iron vire ) which servod to he brass electroc " " \ldquo ; was ( .
joint which copper 1eadsv .
erted .
The ends of these leads , close to the electrode " " wele joined by a short piece of 10 sten wire .
receptaclc filled with mercury , which was ently heated with a Bunsen flame .
One lead to the filament earthed , and the ring was raised to any desired potential above the filament .
A current of amperes the filam erved to heat it a lowing white .
The stl.iking of the arc and its appeared to depend on three things : the temperature of the filament which supplied the electrons , the density of the mercury vapour bombarded by these electron , , and the voltage between the ring and the filament .
Since , when the mercury vapour was made very dense a great deal was deposited about the tubes of the lamp threatened even to enter the spectrograph , very little heat was ] to the mercury after the lamp was once started .
This resulted in an arc which not very , and which failed to at all below ) ever , there appeared to be some conlIsation with a weak the cloud of } ) through which the radiat , ion had to on its through the tube " " \ldquo ; then not , so dense , and less absorption appeared to result .
A exposure with a lvea]arc brought out the line ] , but is line did not appear a much brighter arc was used .
When iron and copper electrodes were used with the lamp } in fig. 6 , the arc proved very difficult to manage because a potential difference of about 150 volts We'lS required in order to make it strike .
When once it struck , however , the resi tance fell almost to zero , and the intense fused the trodes , either causing them to unite or to drop off .
So far method has ) devised to overcome this dofect .
As a spark was adways Prof. J. C. McLennan and Mr. .
J. between the at a lower voltage when they separated , a motor attached to one terminal , to give it a slight vibratory motion , and so break the contact rapidly and keep up a continual sparking .
It found , however , that this did not yield any lines whatever in the region .
It appears , therefore , that the metal must be incan( scent t yield a radiation rich in the very short wave-le1lgths .
llost perfect results by far were obtained with a carbon arc in a lamp used was also that shown in .
Short pieces of were turned to the size of a lead pencil and inserted in iron receptacles on the ends of the electrodes .
While the potential difference applied here only 110 volts , it was found that , with a suitable variable resistance in the was just as as the iron one , but it did not radiate such heat .
The arc had to be attended cJosely , for , as the carbon burned ' slight difference in the distance between the terminals destroyed it .
lVith volCs applied a much hter and more persistent arc was obtained , but it burned the terminals so rapidly that it was ldoned and the lower voltage used .
VII .
Results .
The plates obtained in the present investigation were carefully up , and the , obtained with the different arcs riven in Table T. The spectrum of , it will be observed , has been measured to , that of iron to , and that of copper to .
In the case of carbon the shortest wave-length observed was at .
It was very difficult to get good plates for printing , and a number of the Saint lines do not appear in the reproductions .
Plate 3 will serve to illustrate the results obtained with the mercury and carbon arcs .
ates for and iron were not suitable for printing , although the line clearly ible on then ] .
( a ) comparison of our ults , obtained with the mercury arc , with those obtained by Lyman and Wolff , shows that a number of -lengths observed by these investigations did not come out on our plate Am those which did not appear are A.U. , , observed by both Lyman and Wolff , A. [ I. , observed by Lyman , and also the pair at , observed by Wolff .
( t ) .
results obtained with the copper , it will bc noticed , were rather , and no wave were observed shorter than ( c ) lines given in Table I for iron were obtained on a single the use of the -cooled lamp shown in hile wave ' and ) In a recent by lIcLennil 1 , of the vacuum arc Schumann Hpectl.um , studied with a tluorite wave-lengths -ere observed which did not collt1 .
out on our ] ) ; few ] omlUOll to ] given in ) IIT .
The ngreement in the eaHuremenfH , il only fair .
Too much reliance must not be placed on results ined i ] if present investigation , for no special effort llade to ) ) tain absolutely metals for use in the arcs , and tinle was not available for refining the ) The best that can be is that the same lines always appear Iiff n A samples of the metal On the and Barl ?
By J. C. .
J. F. T. .
of Toronto .
eceived October ) ) [ .
In view of the probability that ionisation lenlent i connected by the quantum relation with the spectral frequency , it has desirable to nine ( value of this frequency for as many of the elements possible .
lts moreover , is best fouud by the lnembers series in a and then the of the series .
Dunz , in his Dissertation , ives the several members of the series for each of the ents mercury , zinc , and cadmium ; and Lorensel , in his ] ) gives the wave-lengths of a number of the members of the selies magnesium , calcium , and strontium .
, however , while ivin .
as the wave-length of the first member of the , for barium , states that the remaining members of the series are not known .
Starting at this point , we recently investigated the spectrum

