GREATER DISPERSION IN THE EXTREMEULTRAVIOLET

By R. J. LANG AND STANLEY SMITH

ABSTRACT

A Two-Meter Vacuum Spectrograph.-The construction and operation of an opticalvacuum-grating spectrograph capable of accommodating a grating of two meters radius ofcurvature is described. Means are provided for introducing and removing the photographicplate as well as altering the focus and inclination of the grating while the vacuum is main-tained.

Doublet Separations in C .- The following doublets in Fowler's series have been resolvedand the frequency separation measured. (2ri - 3 ) (27r2- 3r) AP = 64 in the second order ofthe grating and 60 in the third order. (27ri-36)(27r 2 -3a) Av= 64.5 in 2nd order. For linesX1335 Av= 65 and lines = 1036 Av=62 showing that since the 1036 line belongs to Ciiprobably 1335 does also.

Doublet Separation in Si iII.-In these series the following doublet is resolved. (3cri-47r)(3o-2-4-r) Av=153.

The need for greater dispersion in the Shumann region has becomeincreasingly evident in recent years especially with the discovery of somany ionised atom series. The complexity of the spectra in this regionis such that it has become imperative to be able to measure wavelengths with a greater accuracy than has usually been the case beforethe lines may be definitely assigned to their proper series. In the presentpaper an account is given of the construction and use of a vacuum-grating spectrograph employing a grating of 192.1 cm radius of curva-ture, and some preliminary results obtained with it.

In a previous paper by one' of us a spectrograph using a grating of100 cm radius was described and the spectra for some twenty elementsrecorded. The present spectrograph, while of course longer, has beenconstructed somewhat along the same lines as the former but includesseveral novel features which have made its use possible and thereforeit seems worth while to publish a short description of the new instru-ment before proceeding to record the measurements which have alreadybeen made by means of it.

A horizontal plan of the instrument is shown in Fig. 1 while Fig. 2shows a vertical elevation. The main body A consists of a seamlessbrass tube 20.3 cm in diameter and 225 'cm long. The grating mountingG, the plate glass window R, the large opening L for access to the grat-

IR. J. Lang, Trans. Roy. Soc., A, 224, pp. 371-419;

523

R. J. LANG AND STANLEY SMITH [J.O.S.A. & R.S.I.,12

ing and the opening K to the rough vacuum pump are all essentiallythe same as in the previous spectrograph and therefore require nofurther description.

The end of the tube remote from the grating is enclosed by a specialbrass casting the form of which is illustrated to the best advantage inFig. 1. The source may be a high-potential spark or an arc as desired

and is formed in a separate chamber consisting of a circular brass tube12.5 cm in diameter and 20 cm long with its axis set perpendicular tothe axis of the main body as shown by the dotted circle with center at 0.

FIG. 1

This chamber has bakelite ends which serve for insulation and is

attached to the spectrograph by the conical ground joint J which pro-jects into the source chamber as shown. A large metal tap, formed in thesolid brass casting at N, when open allows the radiation to pass freely

LI~~~~~~~~~~~~~~~~~~~~~~~~~IP

A H

FIG. 2

from the source at 0 to the grating surface, and when closed effectually

separates the source-chamber from the main body of the spectrographfor vacuum purposes. This metal tap is made exactly on the principleof a large glass tap except that the small end (lower end) of the barrelis closed to the atmosphere. The bore of this tap is 1.27 cm.

The plate-holder P holds the plate in a vertical position exactly thesame distance from the grating as the slit which is situated in the end of

the ground joint J at S. The slit and the lines on the grating are hori -

zontal. This plate-holder is part of a tapered steel plug carefully ground

into a tapered opening in the casting so that when the plate-holder is

524

,:I

VACUUM SPECTROGRAPH

inserted into the spectrograph the plug makes a perfectly gas-tightstopper for the opening. Two plates are accommodated one on eachside of the holder, the interchange being effected while the vacuum ismaintained simply by rotating the plug through 180°. The windows Hcoincide with openings in the plate-holder and may be used for roughlyfocussing the grating.

Hitherto many have thought it next to impossible to evacuate avolume of 70 liters such as this and maintain it at the very low pressurenecessary to the production of the high-potential spark. We have foundit quite possible, however, by the use of a General Electric two-stage,rotary, oil-sealed pump, as a rough pump attached at K, to reduce thepressure in half an hour to such a degree that the Kurth diffusion pump,attached to the source-chamber at M, and backed by a Hyvac mechani-cal pump, would begin to operate. The valve at K is then closed and theevacuation continued by the diffusion pump train alone. We havefound, of course, that when the instrument is first exhausted it requiressome time to reach a stage at which the spark may be used, and that thelonger the instrument can be used without admitting air the better thevacuum becomes each time the instrument is operated. On this accountit was realized several years ago by the first-mentioned author, whodesigned and built the present apparatus, that if some method could bedevised for introducing and removing the photographic plate fromthe spectrograph without admitting air there then seemed to be noinsurmountable obstacle to the use of a grating of several meters radiusthus obtaining as large a dispersion in the extreme ultraviolet as hasalready been obtained in the visible region. Consequently much thoughtwas given to the design of some mechanism by means of which thismight be done and the following is a description of the apparatus whichhas successfully accomplished this end.

The lower part of Fig. 3 is an end-on view of the spectrograph propershowing the plate-holder end, and the upper part of the figure shows thedevice for introducing and extracting plates from the spectrographwithout admitting air. The latter consists essentially of the buffer-chamber S 10.8 cm in diameter, and the use of two exactly similarplate-holders X and Y. The method of procedure is as follows: Atthe lower end of the figure the holder Y is shown in position in thespectrograph. When the plates on this holder are to be removed thebuffer-chamber S, already containing the other holder X with its twounexposed plates, is placed in a vertical position with its base in the widegroove in the brass collar shown at B. The bottom of the groove and

May, 19263 52$

R. J. LANG AND STANLEY SMITH [J.O.S.A. & R.S.I.,12

the lower end of S have been carefully machined and ground to fit sothat a little vacuum wax on the surfaces is all that is needed to producea vacuum tight joint, thanks to the large atmospheric pressure forcing

these faces together. S is now evacuated by meansof the Hyvac pump which is also permanently con-

<l |nected at Q. The volume of S is about six litersand the pressure is sufficiently reduced in about fif-teen minutes. The holder Y is now drawn to the

AL, top of the chamber S as shown by means of thehalf-inch steel rod which may be screwed into itsupper end. The holder X is now swung into positionby the eccentric, attached to the second steel rod,and pushed down into the opening in the spec-trograph originally occupied by Y. The air is nowadmitted to S by the tap t and the chamber re-

x moved, whereupon the holder X may be withdrawnand its plates developed. The possibility of leaks at

s the openings for the sliding rods at the top is takencare of by making ordinary steam stuffing-boxes

Y2 around each and covering the same with a smallB vessel full of vacuum oil such as is used in all ordin-

Q t < ary vacuum pumps.This arrangement has been found to work per-

fectly. The usual time required to exchange platesis about half-an-hour. We do not go to the troubleof evacuating S to a very high degree by the use

FIG. 3 of a diffusion pump as we desire only that, after theoperation is completed, the vacuum in the main spectrograph may bestill sufficiently good to allow the diffusion pump to commence oper-ations without any preliminary pumping with a rough pump. Thisis always attained. In fact it is quite possible to preserve a soft x-rayvacuum throughout the operation. This, of course, is because of thefact that, while the vacuum in S is not so high as in the spectrographchamber when the holder is extracted, the volume of S is so much thesmaller that the pressure in the spectrograph chamber is not greatlyaltered.

The metal tap J makes it possible to change the electrodes byadmitting air only to the source-chamber which is quite small and theattachment at Z allows the focus and also the tilt or inclination of thegrating to-be altered while the high vacuum is maintained.

526

VACUUM SPECTROGRAPH

It must be borne in mind that the chief advantage in the use of thisdevice for extracting plates does not consist primarily in the gain intime by not having to exhaust the spectrograph for each exposure,great as that is, but rather in the fact that the spectrograph may bekept under high vacuum for weeks at a time so that all small gas-pockets or gas-traps which continually seep into such a volume, to thedespair of the operator, have long since been exhausted and also the gasgiven off by the metal surface has been reduced to a minimum.

MEASUREMENTS

In 1924 when Fowler published his series in singly-ionised carbon2 the

dispersion at his disposal did not enable him to resolve the pairs ofdoublets in the Shumann region. His calculated value for the constantfrequency separation of the doublets in the First Sharp and FirstDiffuse Series was given as Av = 58.

We have now been able to measure directly this separation for thefirst pair of each of these series in the second and third order of ourgrating. For the first pair of the First Sharp Series (X=858) we findthat Av = 64 using the second order and Av = 60 in the third order. Thisseparation was measured on nine plates in the former case and three inthe latter using a comparator reading to 0.001 mm and taking the aver-age of ten settings on each component. The first pair of the First DiffuseSeries (X = 687) was measured only in the second order and on twoplates. We find for this line that Av = 64.5.

We have also measured the separation of the two components of1335 and also of those at 1036A. Fowler records the latter as a memberof a combination series belonging to Ci but has not included 1335 amonghis Ci lines. Millikan3 has, however, found that the doublet separationof each of these pairs is the same and hence assigns 1335 to C1. Our

measurements would appear to confirm this since we find that for 1335Av = 65 and for 1036 Av = 62. These measurements were made on seven

plates using the second order.We have also made a few measurements on the doublets of Si iV4.

Quite a number of these doublets have not been resolved and shouldfall within the possibilities of our instrument but in most of the casestried these lines are masked by other denser lines on the plates and sofar we have not been able to bring them out clearly. We have, however,

2Fowler, Proc. Roy. Soc., March, 1924.3 Bowen and Millikan, Phys. Rev., September, 1924.4 Fowler, Proc. Roy. Soc., June, 1923.

527.May, 1926]

R. J. LANG AND STANLEY SMITH [J.O.S.A. & R.S.I.,12

measured the separation of the components of the second member ofthe First Principal Series (X =457) for which we find Av = 153.

Since the foregoing was written, we have obtained another gratingfrom Professor Wood of Johns Hopkins ruled with 30,000 lines perinch and having a radius of six feet. The dispersion of this grating isabout 4.5 A per mm in the first order which is slightly less than twicethat of the former. The shortest wavelength registered by this gratingwas about 570 A and below 1000 A it was not nearly so effective as theprevious ruling. It is, however, an excellent instrument for longerwaves and we have succeded in repeating some of our former measure-ments with it especially on the carbon doublets at X = 1335 and X = 1036.We find excellent agreement for these with our previous measurementobtaining as an average value for the former pair AP=65.65 in thefirst order and AP'=65.7 in the second order and for the latter pairAP = 62.6 in the second order and L = 62.6 in the third order which isequivalent to the twelfth order of a one-meter grating with 14,000lines per inch.

The decision reached above, that these pairs belong to the sameionisation of carbon must therefore be reversed. It appears clear thatthe difference in AP, while small is quite real and definite.

It should be noted that our value for the pair at 1036 lies about mid-way between Simeon's value of 58 and Millikan and Bowen's value of66.76.

On account of the lack of effectiveness of this grating for short wave-lengths we were unable to obtain a measure of the separation of thepairs of carbon doublets at 858 and 687.

DEPARTMENT OF PHYSICS,UNIVERSITY OF ALBERTA,

EDMONTON, ALBERTA, CANADA.

528