]\gt ; in England for the purpose of securing a good focus over the greatest possible range of spectrum .
In these preliminary trials the prismatic camera was mounted in front of a large reflecting telescope , the eyepiece of which was replaced by a slit receiving light from a clear sky .
All the spectrum photographs obtained in this way are very dense in the region near compared with other parts of the spectrum , and those which are correctly exposed for give only a very feeble impression of the spectrum in the ultra-violet above K. A very long exposure gave an extension nearly as great as in the eclipse negatives , but the image is extremely feeble at 3441 , whilst the whole of the lowel spectrum from 3700 down to is very greatly over-exposed .
In the eclipse negatives , on the other hand , there is oniy a slight indication of over-exposure at , and yet the lines are strongly shown up bo the limits of the plates , one of which extends to .
( Compare fig. 2 with figs. 1 and 3 in Plate 10 .
) The same kinds of plates were used in both cases , and the very satisfactory results obtained at the eclipse are probably to be attributed to the extreme dryness of the air in Central India at the time of the eclipse .
The following table gives the approximate exposure times , the plates used , and the limits to which the spectra can be traced in the ten eclipse negatives .
As I was working without any assistance , the times of exposure could only be very roughly estimated .
In the case of the four cusp spectra , the times have been calculated from measures of the width of the strip of continuous spectrum .
All the expusures made yielded good negatives , and Nos. 1 , 2 , 3 , 4 , and 7 were selected for measurement .
The first five spectra of the series are nearly perfect in focus between the limits and , but the remaining five , exposed after mideclipse , are not quite so satisfactory , the being good only in the extreme ultraviolet , from 370 to 340 .
O.wing to the imperfect chromatic correction of the lens used , it was found necessary to incline the plate-holder about 7 degrees from the normal to the axis of the instruOF SPECTRA AT THE SOLAR LCLIPSE OF JANUARy 22 , 1898 .
meant in order to obtain a good focus over a long range of spectrum , and the shift of the whole spectrum in the direction of its length , which occurred when the chromosphere arcs changed from the east side of the sun to the west , was to spoil the focus for most of the spectrum , and shift the region of good focus further towards the more refrangible end than had been allowed for .
It was found possible , however , to make fairly good measures of all the lines shown in No. 7 spectrum .
.otographs the Cusp before and 2 Plate .
-The first two photographs of the series are mages of the cusp spectrum , the exposures being made 20 and 10 seconds before totality respectively .
At the moment when the first exposure was made , the ) of photosphere still uncovered had a width of at the centre of the cusp , and this had diminished to only when No. 2 was exposed .
The cusp therefore acted the part of an exceedingly fine slit , and gave beautiful images of the Fraunhofer lines .
The two images obtained may be considered together as they are alike in every respect , excepting in the width of the continuous spectrum .
This is millims. in No. 1 , and millims. in No. 2 , the moon 's diameter on the plate measuring millims. The continuous spectrum shows most of the features of the ordinary Fraunhofer spectrum , the dark lines being represented by very sharply deiined curved arcs .
These are , however , very much less dark than the lines in the spectrum of sunlight , and the hydrogen lines , and S are not present as dark at all .
and are broad and dark , with an irregular reversal at one point showing motion in the lin of sight .
With the exception of the absent hydrogen lines the relative bensities of the dark lines in these spectra appear to be identically the same as in the Fraunhofer spectrum throughout the whole region photographed , but the cusp spectra have pale " " washed-out\ldquo ; appearance when placed beside photographs of the spectrum of ordinary daylight , this peculiarity does not seem to be due to any accident of exposure or developmen ( Compare the dark lines in ] and : ; , Plate 10 , with the corresponding lines in fig. 2 .
) [ After returning to ngland a set of photographs of the Fraunhofer spectrnm was ) the same prismatic camera , for oomparison with the cusp spectra .
These were ) in the manner as the preliminary trial plates mentioned above , a reflecting collimator and slit .
In order to reproduce as closely as ssible the form of the cusp a circular slit was used , formed by a hole cnt in a plate of 1lea ] lie covered disc of meter .
The ) made to snbtend Ilgle of the isms , and the atio of the of tho holo lisc the th of the tIld m at the eclipsc .
In the ) the slit simply directed to sky il OllS 0 Plate 10 , is a reproduction of one of the spectra so ined , in which the of the ) is the as that of the cusp spectrum in fig. 1 .
ngular width of the slit was therefore the same as that of the solar cusp at the moment when the cusp spectnlm was photographed at the middle point).\mdash ; , 1901 .
The bright radiations of the chromosphere are in no cases strong enough to show ( as bright lines upon the continuous spectrum , although they are strongly impressed along the borders , where they form in many cases a continuation of the dark arcs of the Fraunhofer spectrum .
The very strong bright arcs , and , can be traced round the limb of the moon for a greater distance than any others .
No. 2 spectrum appears to have been under-exposed and the bright arcs are rather difficult to detect upon it , although the exposure was made within 10 seconds of second contact .
Measures were not made of the bright lines in either spectrum as they are not very well adapted for accurate determinations .
All of them can be identified with the stronger lines in No. 3 spectrum .
Methods of Reduction of the Spectra .
The spectrum was carefully measured in both cusp spectra and identified line by line with HIGGS'S photographic map of the normal solar spectrum .
As these measures form the basis for the determination of wave-lengths in spectra No. 3 ( flash ) and No. 4 , it is desirable to give a brief account of the methods adopted .
I am indebted to Dr. RAMBApT for the loan of an excellent millimetre micrometer belonging to the Radcliffe Observatory .
This instrument consists of a microscope mounted on a sliding frame , and moved by a long screw with a pitch of 1 millim. The head of the screw is divided into 100 parts , and by estimation can be read to 001 millim. All the measures were made upon this micrometer .
Fach spectrum was measured completely three times over , using different sections of the screw and different readings of the head each time .
By this means any systematic differences due to errors in the screw were eliminated in the mean result , and accidental mistakes in reading the divisions readily found and corrected .
No evidence of any systematic error in the screw was , however , detected , and the three sets of measures of spectra Nos. 1 and 2 ) owed a very satisfactory accordance throughout the whole length measured .
In measuling a line , the mean of three settings of the spider lines taken as the scale reading of the line , and in the final mean result , in which the two cusp spectra are combined , every represents eighteen settings of the spider lines .
measures of the two cusp spectra when compared were found to agree as closely as two separate measures of either spectrum , thus proving that no change had occurred in the dispersion from any cause .
The two spectra were therefore considered to be idontical , and the measures were combined .
the final mean of the six sets of measures about -defined and isolated lin es were selected as standards ; these are all distributed through the ultra- violet portion of the spectrum more than H. In the region between OF SPECTPA AT THE SOLAR 1898 .
and about twenty-six lines were used , which are , however , not so well defined or so satisfactory as standards as the others .
In constructing a scale of wave-lengths for the reduction of the bright-line spectra ( Nos. 3 and 4 ) , the best defined known lines between and in these were also selected as standards , and used together with the twenty-six cusp spectrum lines .
But in the ultra-violet the scale entirely upon the standards of the cusp spectrum , the whole of the lines in this region in Nos. and 4 ( including the hydrogen lines ) being treated as unknown .
A careful comparison made between measures of known lines in Nos. 3 and 4 spectra and the mean measures of the two cusp spectra shows that it is safe to assume that all the spectra obtained before mideclipse may be considered to be identical as regards dispersion .
A sufficient number of standards well distributed throughout the entire spectrum being thus obtained , the approximate relation between scale reading and wave-length at all points in the spectrum was next determined by graphical methods , and from the interpolation curve obtained it was easy to compute the scale reading corresponding to a number of definite points in the spectrum separated by equal intervals of wavelength .
The scale reading corresponding to each 50 tenth-metres of wave-length was in this way computed , taking in each case the mean esult given by three or four standard lines situated near ; and the values thus obtained were -ally slightly COl.rected by smoothing differences until fifth differences were made to increase in a egular progression .
The smoothed values in no case differ from the values derived the measures by quantities greater than the uncel.tainties of the measures themselves .
The mean difference in the best portion of the spectrum to 3400 is millinl .
, and the greatest millim. , the latter corresponding to 15 tenth-metre at 4100 and07 tenth-metre at 3400 .
In the visible the definition under the microscope is very pool in the negatives except No. 4 , and the measures in consequence are not very consistent , for the sake of completeness line has been measured , and an accurate scaale of waye-lengths constructed the entire range of spectrum ] ) hotographed .
From the accordance between the different sets of measures of the bright lines in No. spectrum the probable errors of the wave-length determinations given in Table I. ( p. 404 et seq. ) would appear to be about15 tenth-metre at 4000 , to 05 tenth-metre at 3400 .
-line No. 3 ( Plate 11 , fig. 1).\mdash ; This negative was exposed at the lnent when the lines of the Fraunhofel spectl.um were seen to } ) and innumerable bright lines flashed out across the strip of continuous spectrum which still remained visible .
VOL. CXCVII .
\mdash ; A. 3 .
J. SHED ON -LENGTH IINATIONS , ETC .
, I was observing with a slitless oscope attached to a 3-inch equatorial , and made the exposure as neal .
lie as I could judge 2 seconds before the last trace of continuous spectrum vanished .
The plate was given an exposure of at least 4 seconds in order to impress the fainter details .
In the photograph the strip of continuous spectrum is divided into four parts by longitudinal dark bands due to ities in the moon 's limb .
Most of the bright lines of the chromosphere and flash spectra are visible as bright lines upon the continuous spectrum , but the helium lines at and are an exception , and cannot traced eyen in the dark spaces between the strips of continuous ectrum , although strongly impressed on either side .
The helium lin at 4472 is visible , but weaker in the intermediate dark spaces .
No trace of any absorption lines can be detected in any part of the continuous spectrum .
The total depth of the chromosphere as indicated by the calcium lines and is , these lines extending over an arc of 124 degrees .
The hydrogen lines , S , and the titanium lines at , and , as well as the helium line extend oyer an arc of about 102 degrees , implying a depth of The bright arcs of the flash spectrum proper are approximately of the same length .
On the north side of the continuous- spectrum they are cut off abruptly by a dark band , due to a projecting lunar mountain , and two similar bands interrupt the lines on the south side .
Most of the fainter flash lines can be traced oyer an arc of 40 degrees of the sun 's limb , indicating a depth of , or nearly 600 miles .
A great many of the stronger lines , however ( chiefly titanium , iron , and magnesium lines ) , extend faintly far beyond this limit , and can be traced over an arc of nearly 70 grees , the depth implied in this case exceeding 4 or 1800 miles .
* A few very faint lines due to the very lowest layers of the chromosphere are also indicated in the best defined portion of the negative .
These are visible only upon the strips of continuous spectrum , and .
do not extend beyond .
obably only a small proportion of the finer lines which were actually present at second contact are shown in this ) ) .
In the green region near this is cel.tainly the case .
The striking feature seen when observing the spectrum under high dispel.sion and without any slit was the immense number of excessively fine bright arcs which in a short section of the spectrum ( including the lines ) at the moment when the exposure was made .
Yet very few of these lines appear on the photograph , pal.tly perhaps on account of the poor focus in this region .
[ In estimaling depths from the lengths of the arcs the moon 's apparent semi-diameter is taken at and of the SU11 97 9 .
A uniform distriI ) ntion of the gascs round the limb is of course assumed , and the results from No. 3 spectrum indicate the depth sured from the photosphere to the upper limit to which each be .
The depth of left uncovered the moon this photograph being quite inappreciable , the limb of the moon may be considered as oing coincidont with that of the at the of sscond contact.\mdash ; Mnrch 9 , 1901 .
] OF SPECTRA AT THE SOLAlt ECLIPSE OF JANUARy , 1898 .
The continuous spectrum of the corona is strongly marked on this photograpI ] , but only one true corona line visible , the green line at 5303 .
This line can be traced only on the eastern portion of the moon 's limb , and , unlike the chromosphere arcs , it very variable in inteI ) sity , being strongly marked bout the angles to , and , where the continuous spectrum is also strong , but extremely weak or entirely absent in other parts of the limb .
The results of measures of the corona line on this plate and No. 7 spectra given on p. 401 .
[ In the reproductions of the bright-line spectra shown in Plate 11 , the position angles may be inferl.ed from the fact that the direction of dispersion is also the direction of the moon 's path across the sun , the point of second contact being at the centre of the arcs on the right-hand side , and third contact at the centre the left .
The angles are reckoned from the north point through east , south , and west .
The edge of each spectrum , therefore , represents the south-east limb , and the lower edge the north-west limb .
The diagram belo , showing the prominences in their correct positions , will make this clear .
The broken lines indicate the positions where the green corona line is strongest on the negatives .
In the reproductions it can be faintly traced on the east limb , where it persists throughout totality , thus proving it to be a true coronal radiation , since the chromosphere on this limb was entirely hidden by the moon after mid-eclipse .
In fig. 5 the interrupted coronal arc is perhaps best seen ; it almost meets the oppositely curved chromosphere arc The wave-length scale at the top of Plate 11 must be understood as applying only to fig. 1 , for which it was constructed .
The designation of the ] ) incipal lines in the spectra are given underneath fig. 5 , at the bottom of the ) late .
MR. J. EyERSHED ON WAVE-LENGTH DETERMINATIONS , For further facilitating references to the bright lines a plate is given of the ultra- violet region of the spectrum of the lower chromosphere .
In this a waye-length scale is given above and the designation of some of the principal lines below .
The plate has been prepared by purely automatic methods from the original negative taken at second contact .
By photographing a longitudinal section of this negative with a cylindrical lens interposed , the short sections of the chromosphere arcs are spread out into straight lines .
Great care was taken to exclude spurious lines due to defects in the film or to dust on the slit .
By using a rather wide slit these imperfections have been reduced to a minimum , although this involved a slight sacrifice of definition and loss of detail.\mdash ; March 9 , 1901 .
] Spectrum No. 4 ( Plate 11 , fig. 2).\mdash ; The exposure in this photograph was probably less tham half a second dul.ation , and was made several seconds after closing No. 3 exposure .
The continuous spectrum of the corona is faintly impressed* and the green line is just traceable at the points where it is strongest in No. 3 spectrum .
The chromosphere arcs and and the hydrogen series are narrow and well defined , the latter both in the visible spectrum and the ultra-violet .
The three titanium lines at , and are shown as exceedingly fine threads , extending over an arc of about 60 degrees ; they are distinctly stronger than the line but fainter than .
The hydrogen lines , S and the calcium lines and can be traced nearly the same extent of the limb as in No. 3 spectrum .
The prominences are well defined in this negative and their spectra exhibit some interesting features .
There are seven prominences clearly shown , which I have esignated by the letters of the alphabet in the order of their position angles , as follows:\mdash ; All these prominences are about equally intense in the calcium lines and ; is the largest and and are the smallest images .
* In preparing a plate suitable for reproduction it was found necessary to intensify the image with nranium .
This has made the continuous spectrum of the corona appear very much stronger than it is in the original negativs .
inner corona ; it is most intense at position angles to and to On the west side it can be traced at about position angle , but is extremely faint here .
No trace of ally true corona line can be made out in the blue or violet part of the spectruln , but there is a fairly distinct impression of a line near the end of the plate at 3388 , faint maxima of intensity are indicated on the east limb at the same position angles as the maxima of the green line .
The chromosphere in this negative is represented by the lines and only , which form arcs extending over the western limb .
The prominence gives the rincipal hydrogen , helium , and titanium , lines in addition to and K. No. 6 .\mdash ; The exposure in this case was of 10 seconds ' duration , beginning about 2 seconds after closing No. 5 .
The chromosphere arcs have reappeared in the hydrogen and helium lines on the west limb , and the former can be traced to .
The three strongest titanium lines are also visible , extending over an arc of The prominence shows the streak of continuous spectrum starting at 3668 and extending as a fine thread far into the ultra-violet .
The green corona line is faintly impressed both on the east and west limbs , but no other corona line can be made out with any certainty .
There is , however , a doubtful impression of a line extending pretty uniformly over the west limb at a point in the ctrum a little less refrangible than 398/ 399 .
It cannot be traced at all on the east , at the points where the green line is most intense , and if it is a real corona line , the substance producing it is evidently not the same as that which gives the green line .
Spectrum No. 7 ( Plate 11 , fig. 5).\mdash ; The exposure of 20 seconds ' duration was started immediately after the last and ended only 3 or 4 seconds before the photosphere appeared .
In the chromosphere spectrum is well shown , and in the ultra-yiolet the hydrogen lines can be clearly traced to .
Many of the stronger flash-spectrum lines are visible along the central portion of the photograph extending over arcs of from to The wave-lengths deduced from measures of these lines are given in Table I. , and in the region between and , where the focus is good , the results are as accurate as those obtained from No. 3 spectrum .
A conspicuous feature in this photograph is the shaded band of continuous spectrum in which the more refrangible lines of the flash spectrum are immersed .
It is absent in the region less refrangible than 3668 , and corresponds to the fine streak or tail hich the prominence shows in all the photographs obtained during totality .
The green corona line is better defined in this photograph than in any of the others .
The positions of maximum intensity correspond precisely with the bands of continuous spectrum due to the brightest regions of the inner corona .
The position angles me ) sured on this plate are as follows:\mdash ; OF SPECTRA AT { SOLAR JANUARY 22 , As in No. 6 spectrum there is an exceedingly faint impression of a corona line near and extending over the west limb , but excepting this no trace of any other lines are visible .
The unfortunately , is too feeble in the extreme nltra-violet to show the line at 3388 found on No. 5 No. 8.\mdash ; This negative was exposed for about half a second almost at the moment of third contact .
The reappearing has been impressed as four very narrow streaks of continuous spectrum .
The flash spectrum is well developed , and the whole length of the spectrum is crowded with lines extending between and across the strips of continuous spectrum .
The majority of the lines due to the lowel chromosphere extend over an arc of , the depth implied in this case being The focus in this negative is unfortunately very poor throughout the ctrum , and no measurements were made for determining waye-lengths .
The was , howeyer , carefully compared with No. 3 by means of enlargements of each spectrum , which were made to correspond approximately in scale .
Superposing upon the other it was found that the two appeared to be identical in all respects .
If any differences exist at all between the spectra of the east and west limbs of the sun they can only be found in the finer details , which in No. 8 are lost by reason of the imperfect focus .
Spcctra Spectra Plate .\mdash ; These were given all exposure of less than half a second at about 10 and 18 seconds after contact respectively .
They are both very similar to the first two spectra of the series , but rently the exposures were appreciably of longer duration than those made before totality , and the ight lines bordering the colltinuous spectrum of the photosphere are in consequence much more strongly marked than in Nos. 1 and 2 .
As in all the spectra of the west limb the focus is imperfect except in the ultraviolet region more refrangible than .
The ight lines are , however , fairly well defined along the south edge of the cusp .
The Fraunhofer spectrum is 1tot nearly so well developed in these as it is in Nos. 1 and 2 .
This may be due to wcll ) focus , ) the ultra-violet end where the ) ectrum is less dense the focus good , the dark lines , No figure is given of this it found impossible to produce a ctory plate suitable for reproduction .
although arly visible , are llot nearly so stroI as in the corresponding negatiyes of the east limb before totality .
All the more conspicuous lines which characteristic of the flash spectrum can be uisbedD along the south edge of the cusp in both No. 9 and No. 10 spectra .
GENERAL RESULTS CONCLUSIONS .
The sh Spectrum .
In comparing the wave-length values of No. ; and No. 7 spectra given in Table I. with ROWLAND'S wave-lengths of the absorption lines in the solar spectrum , it is at once apparent that practically eyery strong line in the latter is present in the lower chromosphere as a bright line .
In the region between ) and 4410 there are 58 strong dark lines with an intensity exceeding 8 on ROWLAND'S scale of intensities .
In the flash spectrum ( No. 3 photograph ) 44 of these lines are certainly present as bright lines ; 6 are probably present , 6 are too near to strong hydrogen lines to be separately distinguishable , and 1 is by H. One lin only of the 58 is unaccountably absent , that at ( Fe , intensity 9 ) .
It is further to be observed that of the bright lines of the lower chromosphere ( hydrogen and helium lines bein excluded ) the great majority appear to be coincident with dark lines having an intensity on ROWLAND'S scale not less than 3 , and two lines only and occur in a blank space in the solar spectrum where the lines are weaker than intensity It may be said generally therefore , with regard to the ultra-violet region , that the bright lines of the flash spectrum are reversals of Fraunhofer lines , including the very strong lines ( intensity 10 and upwards ) .
Whilst the positions of almost all the bright lines appear to coincide with dark lines in the solar spectrum , the relative intensities of the lines in the latter are widely departed from in the flash spectrum .
A negative of No. 3 spectrum is therefore unlike a positive of the solar spectrum .
When , however , the flash spectrum is analysed into its separate constituents.it is found that the relative intensities of the lines of any one element correspond very closely with the intensities of the dark lines of that element .
In the iroll and titanium spectra , which claim more lines of the flash spectrum than any other elements , this correspondence is most shown .
There are , as might be expected , many cases where the intensities are abnormal , but by taking the average intensity on ROWLAND'S scale of all the dark lines corresponding to each unit of intensity in the flash spectrum the effects of these are eliminated .
These abnorlnal intensities are probably due to the superposition of closely adjacent lines .
The following tables prove the general close correspondence of intensities in the flash and solar spectrum for and titanium .
The first column gives the number of flash lines having the estimated intensity given in colunm 2 , whilst the average OF SPECTRA AT TH SOLAR ECLIPSE OF JANUARY 22 , 1898 .
It thus appears that the tensities found in these spectra by no means represent the actual intrinsic intensities of the bright lines of the different elements .
Conld we obtain a sample of the incandescent gas from near the base of the flash-spectrum layer and examine it close at hand with a slit spectroscope , it is certain that the relative intensities would differ widely from those in the flash spectrum as observed at eclipse , and it is possible that they would be found to correspond much more closely with the relative intensities in the Fraunhofer spectrum .
From the foregoing considerations it is clear that the emission lines from the lowest levels of the ilash layer must be very difficult to obserye , however bright they may be intrinsically .
The wide divergence between the flash and Fraunhofer spectra , with respect to intensities , would appear , therefore , to afford no ground for abandoning the original interpretation of the flash proposed by Professor YOUNG from his rvation in 1870 , and the evidence of these photographs certainly indicates that the flash does , in fact , represent the upper , diffused portion of a true reversing stratum .
In the flash-spectrum photograph ( No. 3 ) fifteen elements can be identified with certainty in the lower chromosphere , in addition to hydrogen and helium , and there are three elements doubtfully represented .
Arranged according to their relative intensities , the four groups occur:\mdash ; .
J. ON WAVE-LENGTH DETERMINATIONS , ETC .
, Group IV .
( Lines weak in flash and in spectrum .
) . . .
atomic weight . . .
, , , 12.0 The elements of which the identifications are somewhat doubtful are\mdash ; Si atomic weight One line only Co , , , , Mally Co lines seem unaccountably absent in the flash La . . .
, , , , Two lines only It is probable that the elements of Group I. are diffused throughout the entire depth of the flash-spectrum layer , but become denser near the photosphere .
Those of Group II .
are possibly absent from the very lowest strata , but are widely diffused in the higher regions ; whilst the elements Fe and Ni ( Group III .
) would appear to be mostly concentrated in the lowest strata where the density is great enough to produce the winged absorption lines .
Group IV .
probably represents low-lying elements of small density .
The bright lines of the flash spectrum corresponding to Group II .
may be considered to be true reversals of the dark solar lines of these elements ; that is to say , the whole of the matter concerned in the absorption contributes to the emissiori spectra , and the bright and dark lines are practically of equal width .
But in the emission lines of Groups I. and III .
the radiation from the very lowest region , where the density of each element is considerable , contributes very little indeed to the total light , most of which comes from the higher more extensive regions of low density ; consequently the lines appear narrow and without appreciable shading .
These lines , therefore , can only be considered as partial reversals of the corresponding absorption lines .
The metallic elements found in the lower chromosphere ( including those doubtfully identified ) include all the known metals having atomic weights between 20 and 60 , with the single exception of potassium .
Arranged according to their atomic weights , they MR. J. ON WAVL-LENGTH RMINATIONS , identified .
In the egion between S340 and 4000 there are 29 such lines in a of 225 , and all of them are weak lines .
The strongest ( intensity 2 ) correspond in position with dark lines in ROWLAND'S table which have not been identified with any element , and these are given in the following list : These are the only unknown lines in the flash spectrum with an intensity greater than 1 , and it is just possible that they belong to the same element .
The Hydrogen Spectrum .
The waye-lengths of the hydrogen lines taken from Tables I. and II .
are entered separately in Table III .
, together with the values computed from BALMER'S formula .
The latter were computed for a vacuum and corrected for air in accordance with a table of RUNGp Astronomy physics , .
In the formula ' is the series number of the line , and is a constant derived from ROWLAND'S measures of the lines , and in the solar spectrum reduced to a vacuum , the mean value adopted bein In obtaining the mean values given in column 4 , equal weights are given to No. 3 and No. 4 spectra , both of which are in good focus beyond , and give equally consistent measures in this region .
No. 4 spectrum is , however , much the better of the two in the visible region , where the images are comparatively small , well-defined spots .
This spectrum , as already mentioned , consists of a series of images of a very small bright prominence , and very consistent measures were obtained throughout .
The good agreement of the wave-lengths of the metallic lines with values will be seen on referrin to Table II .
The very close agreement of the hydrogen lines with the computed values is shown in column 6 of Table III .
( observation \mdash ; calculation ) .
It will be noticed , however , that the wave-lengths of the more refrangible members of the series are appreciably smaller than the theoretical values , the differences increasing towards the limit of the series .
This may be due to the greater uncertainty of measurement of the fainter lines , which are apt to be confused with other lines closely contiguous , or it may possibly be due to a slight progressive error in the scale value for this region .
The results obtained for No. 7 spectrum given in column 7 appear to confirm the OF SPECTRA AT SOLAR OF JAN TARRY 2 , 1898 .
values of the more refrangible lines in No. 3 , although they were obtained by quite independent methods of reduction .
The measures of this plate , however , cannot be given the same weight as those of the other two , the definition being poor owing to imperfect focus .
The ]ines in the visible spectrum , and S are entered for the sake of completeness , but they are not to be compared in accuracy with the ultra-violet lines .
This is largely due to the difficulty in measuring the broad over-exposed lines and S on No. 3 spectrum , and partly to the small dispersion in this region .
The line S appears in all the measures to be largely displaced towards the red , but this is probably accidental .
It is remarkable , }Jowever , that in table of the solar lines HS is given at , which is also less refrangible than the theoretical position , the difference being tenth-metre .
The almost perfect agreement of the other absorption lines in his table , and the very close agreement of the best defined ultra-violet lines in Nos. 3 and 4 spectra with the positions assigned by the formula , suggests that the wave-length of this line has been erroneously estimated .
Mr. JEWELL , however , who made the measures of ROWLAND'S plates , while admitting some uncertainty in estimating the centre of the line in the solar spectrum , finds on his measures , no justification for altering his original estimate Physical Journal , .
It is very desirable that accurate measures be made of HS in the upper chromosphere where the lin is narrow and free from interfering lines .
The limit of the hydrogen series defined by the formula when is at 3646 .
But in the chromosphere spectra the lines fade away to invisibility long before reaching this point .
I would call particular attention , however , to the very remarkable band of continuous spectrum shown by the prominences and lower chromosphere , beginning near the end of the tabulated series and extending nitely towards the more refrangible end of the spectrum .
The interspaces between the hydrogen lines are quite clear of this continuous spectrum , which begins abruptly at about 3668 .
( See Plate 12 .
) It seems probable that this faint spectrum may be itself due to hydrogen .
In the ab sorption spectrum of white stars of Type I. , Sir WILLIAM has ob , seryed an analogous feature .
This consists in the " " rather sudden fall of ensity of the continuous spectrum at about the place of the end of the series of lines .
The enfeebled spectrum inues to run on without further ) until it is stopped by the absorption of our atmosphere * This seems clearly to be the absorption effect of the matter iving the faint continuous spectrum in my plates , and the fact that the feature is istic of white stars in which the hydrogen absorption is very strong points to its being due to hydrogen itself 'An Atlas of Representave IStellar Spectrit ' ( Sir and Lady ) , OF SPECTRA AT THE SOLAR OF , 1898 .
the emission and absorption lines in the sun , is extremely difficult to reconcile with the ordinary appearance of the lines and at the sun 's limb .
With a radial slit the emission lines are seen to correspond exactly both in width and absolute intensity with the dark lines , and it is difficult to avoid the conclusion that the depth of the chromosphere is effective in producing the absorption .
The progressive diminution of intensity of the emission lines in from the visible spectlum towards the ultra-violet appears to be in itself sufficient to account for the disappearance of ption beyond H. The total quantity of hydrogen above the photosphere is probably too to produce appreciable tion when the lines have a certain limiting intensity , although the same lines when viewed at the sun 's limb may be strikingly conspicuous , chiefly by reason of the wide diffusion of the gas as already explained in the case of the metallic lines in the flash spectrum , and also on account of the enormous of radiating gas which the line of sight passes .
Unlike helium , hydrogen is very conspicuous in the lower chromosphere , and the fainter more refrangible lines are much more strongly impressed in the obtained near the times of the two internal contacts than in those taken lnideclipse with much longer exposures .
The intensity of the lines evidently increases rapidly towards the photosphere , but they still remain narrow and well defined even within of the photosphere .
The Corona Spectruqn .
Although the continuous spectrum of the corona is strongly shown on most of the plates exposed during totality , the green line is the only one of which it was possible to obtain measures .
The faint line near in Spectrum No. 6 becomes quite visible under the microscope , and the new line in the extreme ultra-violet , shown in No. 5 , was not discoyered until some time after making the measures ; the wave-length of this line was estimated by superposing No. 5 spectrum upon No. 3 , the hydrogen and and being made coincident ; the position of the new line was then obtained with reference to the fine lines of the flash spectrum .
The following are the waye-lengths obtained for the green line which was lneasuled on negatives No. 3 and No. 7:\mdash ; * The same may be said of the calcium lines and K. In the upper chromosphcre these linCb c narrow and well defined , and correspond exactly with the rrow d lines the centres of the absorption lines .
It is impossible in the reproductions to show the corona lines here mentioned , except the grcen line .
VOL. CXCVII .
\mdash ; A. .Mli .
J. ON WAVE-LENGTH DETERMINATIONS , ETC .
, This result is in satisfactory agreement with the value found by Mr. FOWLER on plates obtained by Sir NORMAN LOCKYER .
The difference between the two measures above is due to the opposite displacement of the line in the photographs taken at second and third contacts , the point measured being situated about 2 ' above the moon 's limb .
This apparent displacement affects all lines of a diffuse character where the moon 's limb is not well defined .
The three corona lines shown on these plates are therefore\mdash ; 1 .
2 .
Near ( on less refrangible side ) .
3 .
5303.3 .
The ultra-yiolet lin is similar to the green lin in distribution round the limb , and is probably due to the same substance , whilst the line near differs from the others in showing no maxima of intensity , the substance producing it being evenly distributed on the west limb .
There is no trace whatever of any other coronal line on any of the plates .
SUMMARY 0F RESULTS .
The principal results obtained from the study of these photographs are stated briefly in the following paragraphs 1 .
Practically every strong dark line in the solar spectrum is present as a bright line flash spectrum .
2 .
Almost all the flash-spectrum lines ( excepting those due to hydrogen and helium ) coincide with dark lines in the solar spectrum .
3 .
The relative intensities of the lines of any one element in the flash spectrum are practically the same as those of the same element in the solar spectrum .
4 .
The relative intensities between groups of lines belonging to different elements are widely different in the flash and in the solar spectrum .
5 .
The apparent intensity of the radiation from lement in the lower chromosphere is determined by the extent to which that element is diffused above the photosphere , : and the real relative intensities between the different elements cannot be judged in photographs of the flash spectrum .
6 .
The flash spectrum may be considered to represent the upper more extensively diffused portion of a stratum of gas which , by its absorption , giyes the Fraunhofer dark-line spectrum .
7 .
Nearly all of the known tals having atomic weights less than 60 are represented in these spectra , and no element having a higher atomic weight than 92 is certainly represented .
8 .
All the strong lines in the flash spectrum can be identified with known elements , and the small proportion of unidentified lines are weak lines .
OF SPECTRA AT SOLAR ECLIPSE OF JANUARY 22 , 1898 .
9 .
The wave-lengths of the hydrogen lines in the ultra-violet agree closely with those derived from BALMER'S formula .
10 .
The prominences and lower chromosphere emit a strong continuous spectrum in the ultra-violet , beginning near the limit of the hydrogen lines and extending indefinitely in the more refrangible region .
11 .
The ultra-violet hydrogen lines increase in intensity towards the photosphere , but remain narrow lines in the flash-spectrum layer .
12 .
Some of the helium lines decrease in intensity towards the photosphere , others increase .
In have to acknowledge my indebtedness to Dr. RAMBAUT for much yaluable help and advice in the preparation of this paper , and for the kind interest he has taken in the work throughout .
I have also to acknowledge the great assistance rendered to me by .
L. E. JEWELL , of the Johns Hopkins University , who made a careful and exhaustive study of some positive copies of my plates .
In identifying the lines given in the tables , I have been guided almost entirely by his results ; and I am also indebted to him for supplying me with revised values of many of the solar lines giyen in the last column of each table .
TABLE I.\mdash ; Flash Spectrum -lengths with the Dark Lines of thoe Spectrum .
In this table the -lengths deduced from the measures of No. 3 and No. 7 spectra are entered in the first two columns .
The third column gives the intensities for No. 3 spectrum .
These were estimated on a scale ranging from 1 to 10 ; 1 nting very weak lines , and 10 representing the strongest lines in the spectrum .
Very faint traces of lines difficult to measure are designated in this column .
In column 5 the wave-lengths of the solar lines are given with which the bright lines have been identified with a aoreater or less degree of probability .
The intensi- ties as given by ROWLAND are entered in column 6 , and the element corresponding to the solar lin in the last column .
The wave-lengths of the helium lines are placed within brackets ; they are taken from the tables of RUNG and PASCHEN .
The columns 5 , 6 , and 7 are taken from ROWLAND'S tables of solar lines published in the ' Astro-Physical Journal ' for 1895 , 1896 , and 1897 , and supplemented by more recent determinations supplied to me by Mr. JEWELL .
In a few instances the identifications and intensities will be found to differ from those given in the tables .
In column 4 an ( S ) means a sharply defined line , well identified , and suitable to be taken as a standard line in wave-length determinations . ?
in column indicates that the identification with the solar line is extremely doubtful .
In the last column a ?
means that the element in question has not been identified with certainty with the solar line .

