]\gt ; [ 361 ] X. Critical Study of Spectral Series .
\mdash ; Part I V. The Structure of Spark Spectra .
W. M. HICKS , Received , \mdash ; Read , 1916 .
[ PLATES 4-18 .
] THE spectrum of an element produced by the electric spark is in general funda- mentally different from that produced by the arc .
The latter is marked by the predominance of groups of lines forming series associated with one another , and by lines related to these series in definite ways .
In the former very few series have been discovered , the lines are very much more 1lumerous , and the great majority are weak .
Both kinds vary very much according to ths way in which they are produced , and this is specially the case with spectra produced by the spark .
Even when they are produced by similar methods by different observers\mdash ; as , for instance , by EDER and VALENTA , and by EXNER and there will be found a large number in one not included in the other and vice versed .
It is even questionable whether it is possible to draw a distinct and line of demarcation between the two , e.g. , in the cases of Cu and Ba , to take two instances ; the arc spectra\mdash ; with very numerous lines\mdash ; in many respects have analogies with those of the spark .
But very little is known as to the structure of the latter , beyond the fact that in some of them sets of lines with the same frequency differences are met with .
The present communication is an attempt to throw some light on this question .
The material at disposal is so vast that it is necessary to limit the discusslon to a few elements , and even in them to restrict it to certain relations only .
elements selected are silver and gold .
It had ally been intended to include also copper and barium as illustrating the transition from one to the other kind of spectra .
But their inclusion would have rendered the prese1lt communication u1lduly long , whilst their omission enables us to confine the discussion to the elaboration of a single principle .
The reason for the selection is that the author has had occasion to study with special cars for other purposes the spectra of Ag and Ba , whilst those of Au and Cu\mdash ; belonging to the same group as Ag\mdash ; were specially investigated to see if the results afforded by Ag were supported by them .
That of Cu was also interesting , as in the arc there are a very large number of doublets and triplets discovered by RYDBERG ; indeed the origins of some of the frequency differences observed were first VOL. CCXVII . .
8 [ Published October , 1917 .
ested by the case of Cu in which their values are comparatively small .
But it required support of the larger values afforded by Ag and Au to firmly establish the relations .
Similar relations have been found in many other elements .
The spectra of the rare gases from No to RaEm are built on a precisely similar plan , and in fact a 1nap for some of the Kr lines was drawn many years ago on the plan of those given here for and Au , } at that time the origin of the differences was not known .
The notation used is of preyious papers in this series .
The letters , are used to denote the sequences in the alkalies give the variable parts the and Sharp series .
The wave-numbers of the Principal series are then and of the Sharp .
doublet separations are to denominator differences denoted by , say and .
The quantity is a multiple of the ouu , , where is the atomic .
As however the quantity occurs very frequently , this is generally used .
When is added to the denominator in either of these sequences a new line is produced , and it is said to be laterally displaced .
Thus , denoting any line , if is added to the variable part ( the term in ) , the new line is written , but if added to co1lstant term ( the limit ) it is written .
liegarded from this point of the two sets of a doublet series are regarded the one as a displacement of the other .
; In the following , for the sake of sbortness , when are used alone they stand for the limits , .
Line separations are given in thick type , and , in general , the decimals of wave-1lumbers are lnitted .
It will be seen later that the structure of tlLe spectra depends on long series of constant differe1lces in wave-numbers ( or frequencies ) .
As the total number of lines in a particular spectrum is very great there will always be a certain number of chance agreements for any given difference .
In forming an opinion as to the reality of the connection indicated by a given set of such differences it will then be necessary to obtain an estimate of the number of such coincidences to be expected , on the supposition that the lines form a purely chance distribution .
Owing to observation errors exact differences will never occur .
Suppose that the difference being investigated lies between , where is a small quantity , and suppose that the spectrum under consideration contains lines whose wave-numbers lie between and The average separation between successive lines will be .
The chance of one falling in the space is therefore .
Taking each number in succession from the smallest , this is the chance of finding a line ahead of it .
But clearly we must stop when we reach a value .
The probable number of lines contained in this region is and the probable total is reason to belisve , however , that the ries P and the are the normal , and that should be written , \amp ; c. DR. W. M. HICKS : A CRITICAL STUDy OF SPFCTRAL SERIES .
number of lines available for pairs is , since is a very large number .
Hence the probable number of coincidences will be This is the number if we know that the distribution is a wholly chance one .
If , however , the frequencies are known to be based on certain constant differences the numbers found with differences which are not sums or of multiples of these constants will not occur .
Hence , in this case , the calculated probable number will be too large .
It therefore must be regarded as a maximum value , and the chance coincidences will be fewer than the value given by it .
The first difference to be tested was Ilatul.ally that of the ordinary doublet separation .
In the case of Cu a very frequent difference found was about 1000 .
This is rather larger than 4 it gests t is formed by the displacement on , the new difference is formed by the displacement .
But the calculated value of is too great .
value of , however , comes to , which is practically exact .
This is the difference of he two displacements and on , that is of symmetrical displacements on either side of , which latter is probably more fundamental than .
The corresponding values for Ag and Au were also found to occur in large numbers in their respective spectra .
This large difference naturally suggested a search for the intermediate ones , , , which were duly found as well as others given by , There are also certain others depending on seqnence and the satellite separations , but these latter are not considel.ed in the , present paper .
It is to be noted that if , instance , the -number of a line is given by , the addition simply displaces the line to .
If , however , it be subtracted , the llew wave-lulnl)e ) and a nornlal displacement does not occur .
As will be seen later there are numerous instances of displacements of the first kind .
The method adopted in attacking the spectrum of a give element was first to make a list of the wave-numbers in a vertical column of values downwards .
In order to test givcn difference ) in succession was tested , and when a cidence was found pencil lines were drawn down and up with the actual difference found entered .
As a rule values were entered when different that being tested by two or three units or more .
The result was a maze of pencil lines , many of the wave-numbers having as many as five or six attached to them .
The next step was to start from some particular wave-number and make a new list following up the pencil lines from one to another .
In this way a set would be isolated from the general list , all connected together in parallel and series groupings with one another but showing no connection with the other wave-numbers .
In most cases the number of individuals in a set was found to be very large and it is only possible to visualize their mutual connection by exhibiting them on a diagram .
They appeared connected to one another by links of a few definite species .
It will be convenient to refer to these sets as linkages , and a constant difference as a link .
The vast majority of the spark lines are thus connected .
As a rule the linkages are attached related of the ordinary series lines .
It will be convenient for shortness of statement to denote the various links by definite letters .
The following will be used\mdash ; From the constitution of a linkage it is clear that if denote the wave-number of the first\mdash ; or any one\mdash ; of the linkages , that of another can be expressed in the form\mdash ; Wave-number where are integers .
From the way in which the links appear as the difference of two terms the sum must be odd .
The existence of lines in which this sum is even is not excluded , as the method adopted would not isolate them and no attempt has been made to test for them .
Tables for certain of these linkages are given below .
We now proceed to give the evidence on which the above statements are made .
The data for are taken from Part III .
of this series of papers .
* They are :\mdash ; .
Au .
' The values of the different links as calculated from these are as follows:\mdash ; 88 77 92 44 100 26 377 00 245 64 4607.77 * Phil. Trans , vol. 213 .
DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
The exact values of these links depend only slightly on the exactness of the values of the limits , for if the latter change by also changes with it in such a way that the consequent changes in the links are only small .
For instance even with the large values of Au , the links become .
The links , however , depending on a limit which does not affect , show larger , although still small , variations , viz. , and ?
) In the case of Ag , the greatest and least values of the wave-numbers tested are , whilst the total number of lines is .
Substituting these values in the formula given above the number of probable coincidences with a given link is The corresponding values for Au are , whence the probable number of coincidences is .
In order to form a judgment on the question whether the observed coincidences are due to a ) real effect or to mere chance , the series of curves Ag 1-7 ( Plate 4 ) and Au 8-14 ( Plate 5 ) have been .
The actual observed diffel.ences are represented as abscissae , a dot being placed for each case .
For any particular difference , the number of dots within on either side of it are counted and entered as an ordinate .
Joining the tops , we get a broken curve which represents the frequency of occurrence of the various values of an observed link .
A dotted horizontal straight line drawn at a distance- above the line of abscissae , calculated from the formula , gives the probable frequency on the supposition that the values are dne to pure chance .
The curves are drawn for the two cases of .
and 1 .
The curves for the links are made to cover a wider region than the others for a reason considered below .
The link for Au has also been extended further in order to see how the observed differences behave when they are away from the immediate neighbourhood of the true link .
As is seen , their frequency is much less than the calculated chance , in agreement with the reasoning given above that the chance distribution must be less than that given by the formula .
In the Ag set the curves show that the fonr small links , ) , are clearly systematic , with numbers more than twice the maximum probable on chance and practically agreeing in value with the calculated results .
The three larger links also show the same effect with the distinction that they appear to be arranged , about three maxima , roughly two units apart for , and three units for , with itself actually at a minimum .
In Au , the existence of peaks is even more marked .
They are probably all real effects and indicate that certain small displacements will often simultaneously 0CCU1 ' with the chance of link .
The existence of observation errol .
S will flatten out any peaks it they exist in the lrves 1 given above .
An estimate of the magnitude of the effect to be thus expected is difficult , not only because we have estimated maximum errors by KAYSLR and RUNG alone for the arc lines , but because frequently an observed diflerence has to be taken between measurements by two observers whose systematic errors may be different .
The for the reality of the variation in a link is based on the existence of long series of the same link .
The following are a few examples out of many others .
In Silver\mdash ; ( 1 ) 30514 37896 40353 .
( 2 ) 17814 32928 .
( 8 ) 30959 38513 42286 44904 In Gold\mdash ; ( 4 ) 15293 21558 27191 32822 3192 .
3197.75 .
( 5 ) ( 6 ) .
The first three indicate without any doubt that the links about 2460 and 3778 are real , and they show that they cannot be explained by errors of observation , for if so , the extreme lines would have to be affected with errors beyond any possibility .
For stance , in ( 2 ) , the rences refer clearly to one about and individual errors of small amount assigned to the lines will make the differences ffi equal to this .
But if the link is supposed to be the calculated 3771 , the errors in the extreme lines would have to be distributed between them\mdash ; a value manifestly too large .
The existence of these variations is also shown clearly by ( 5 ) ( see p. 367 ) .
The origin of these variations is discussed below in more detail .
The bility curves offer incontestable evidence that the existence of these links is not due to chance coincidences .
But other evidence is available .
The fact that the majority of the spark lines are thus co1lnected can be explained in no other way .
A glance at the maps of the linkages given below , and consideration of the way in which the links are combined , are convincing that although some of the links shown are certainly due to coincidences the linkage effect is a real one .
Amongst such peculiarities may be mentioned the long sequences of the same link referred to above , the existence of numerous meshes and congeries of meshes , of parallel sequences , repetitions , \amp ; c. A few of these are exhibited in diagrammatic form in Plate 6 .
DR. W. .
HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
They are excerpted from the maps of some of the linkages given later .
In these the dots represent wave-numbers of spectral lines , the small circle corresponds to the wave-1lumber which is given in the notes to the figures , and the letters denote the links .
We shall call a collocation of four lines connected in parallel\mdash ; as for example the two sets in fig. mesh .
The number of these is very large and the Au spectrum contains some very complicated examples .
In a very large number of cases a correction on a single line will render all the links attached to it correct .
Thus , for instance , in ( ii .
) and ( iii .
) maps , the line 42203 Plate differs from two preceding lines by and from two succeeding lines by 95962 , .
If this be corrected by ) the differences become the links \mdash ; practically exact .
Another example may be taken from the line 40353 , which occurs in the first of the long sequences given on p. .
In addition to the link there shown , it forms part of a mesh which may be written :\mdash ; 959 .
923.99 919 .
963.29 If 39430 be larger the three links converging on it are made simultaneously exact , , the outstanding variations being only small fractions of possible observation errors .
The link also becomes the same as that just before 40353 in the long sequence to which the abnormal 2460 of the others have returned .
In these two instances the first may be due either to an observation error or to another cause .
The second , however , is too large to be explained as due to an error , but is probably due to the same cause that produces the double peaks in the frequency curves .
That such causes exist has been already pointed out by the many long sequences of equal and abnormal links .
As additional evidence the following examples out of a large number of similar ones may be taken .
In the linkage the following mesh occurs\mdash ; 3192 .
3197.75 Here alterations of on 41320 and on 41325 maks the differences two exact links .
But it would also make the two lines coincide .
whilst they are given clearly by ] and VALENTA as two , 10 , so that any supposition as to the presence of observation errors is ruled out of court .
There is therefore a displacement either in the links themselves or in a line The other example is from silver , in the linkage , 2461.79 .
Here again the two ?
links deviate from their normal values by equal amounts Their mean is , which is exact .
The two lines 31815 , 31809 are 20 and are clearly displacements by equal amounts in opposite directions from The number of cases similar to the above in which a change in one wave-number will make a number of abnormal links all conform to their typical values is extremely large .
When two such links occur in series\mdash ; both in the same direction\mdash ; I refer to them as series inequalities ; when in parallel , or one additive and the other subtractive , as parallel inequalities .
If one is too large and the other too small by equal amounts , it means that when in series , the line between them is displaced from the first one , and the third has this displacement annulled .
If in parallel it means that the third line receives the same displacement over the second that the second received over the first .
If , however , the two links differ from their normal values by the same amount the above are reversed .
In certain cases , especially where linkages are involved , the nature of the displacement would seem obvious \mdash ; where , for instance , the irregularity can be explained by a change from a type to a or vice versed .
Thus in the linkage the line 33192 is greater than the line 30572 .
The former depends ( see No. 7 of the Table ) on .
The satellite separation is by observation .
Thus link change from to is , leaving 89 to be distributed between observations of the two lines and of the satellite separation .
The first line is , the second would then become ( 4 ) .
A similar explanation is possible for the previous example from the linkage .
The line 31815 contains-VS ( 3 ) in its formula .
Now VS ( 3 ) , a displacement of on this produces a separation .
So that numerically 34271 is 31815 , and 31809 is 31815 .
Such explanation , however , is not so plausibl as in the preceding case , for the sequences do not seem to be so susceptibls of small displacements as the D. The most frequently occurring irregularities , however , cannot be explained in this way for the linkages show DR. W. I. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
36 !
) them of the same value as the others , and the smallest displacement of on UP ( 1 ) produces separations of far too high a value .
It is necessary to look for some cause which will produce variations of from 2 to 10 in all cases .
a cause wonld be small variations in N. The condition is also fulfilled if , instead of making the links depend on displacement operations by on , the operations are ) on values of , already displaced by small multiples of or .
For example , , , ( , calculated changes are found to be as follows : \mdash ; Ag ( with ) .
( with ) .
' The refer to multiples of for Ag and of for Au .
The reason for the difference of treatment is that where those of might occur in Ag the change would be much smaller than the observation errors , whilst if any of occurred in Au the changes wonld be larger than those included in our purview .
In constituting the formulae for the lines of certain linkages carried out below this method has been adopted , when the variations are too large to be ascribed to errors .
This assumption , however , must not } ) considered to be adopted as the real explanation .
This is left open .
The use of the assumption enables the dependence the lines on the and terms to be shown conveniently in .
At same time it exhibits clearly the presence of these small disturbances when existing , whatever their real cause may be .
In many cases also after in the tables for certain lines they disappear again for succeeding lines , and so have assisted in the latter .
In the case of the long link , which is the sum of the four short , it might be expected that some dication of the nature of this displacement might be given by considering how the , or the enlarged links depend on the displaced small links .
The number of the ] being large , and the number of possible arrar)gements of the small links within it being 4 !
, the labour of testing them is so great that a complete discussion has not been attempted .
The chief difficulty is that only a few of the small links in them are normal so that the others do not appear in the general lists of links , and the whole spectra have to be examined afresh .
In the case of the and linkages no complete sets were found .
In the of the links beginning at 25736 , 20100 , 37357 , 41172 , 26236 , 43497 ( see map , complete sets were found for 26236 only .
In the of those at 25387 , 20776 , 27783 , 30109 , 24477 see map VOL. CCXVII.\mdash ; A. 3 .
W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
all but the last gave sets , but only by admitting values of links so abnormal that they were not counted in the general list .
Also in the linkage the sets beginning at 17788 , 28152 were found to exist .
They are reproduced here : 1 . . . . . . . . . . .
43497 30067 :34669 40305 2 .
'25387 . . . . . . .
42644 .3 .
20776 . . . . . . . . . .
38038 4 .
2380.5 . . . . . . . . . . . . . . .
41064 27001 30823 .36442 27615 32219 .35414 . . . . . . . . .
41855 27783 .33431 38038 . . . . . . . . . .
45043 . . . . . . . . . . . .
47.361 8 .
17788 . . . . . . . . . .
35045 .
28152 . . . . . . . .
45412 .31967 36578 42216 ( b ) .
28155 . . . . . .
45412 An examination of the above shows that , with the exception of Nos. 8 and 9 from , all contain very abnormal links .
It is noticeable that in all except 3 and 4 the links occur in cyclic order backwards or forwards of ; ( 2 ) ; whilst in ( 3 ) order is , and in ( 4 ) , or .
In ( 1 ) 30067 bafi no link in the general list .
In ( 2 ) 39456 has not been attached to any linkage and the other two belong to the X and linkages .
In ( 3 ) all are included in the .
( 4 ) shows two possible sets both contravening the cyclic rule .
If the links shown real displaced the two sets form a complete cycle like that in fig. ) of Plate But 36442 belongs to the linkage , whilst in the second the three inter1nediate belong to other linkages .
It is probable , therefore , that both sets in No. 4 are spurious .
Nos. 3 , 5 and 6 have portions in common .
In ( 7 ) the last two intermediate belong to the X linkages .
The set is therefore probably spurious .
show all good links ] shows that two links are possible but 31970 DR. W. M. HICKS : A CRITICAL STUDy OF SPECTRAL SERIES .
is an arc line .
Apparently , therefore , there is only good evidence for all the intermediate links in Nos. 1 , 3 , 5 , 6 , 8 and 9 .
In applying the displacement rule to the links in the formulae lists iven below the numerical values allowed are supposed to be and \mdash ; in other words , is taken as an independent link and not as the composite .
If it is composite each of the may be modified , and values of intermediate to 17253 and 17264 may enter .
If then the wave-numbers were known accurately to unity for long waves to 05 for short ) ) ought to serve as tests of the kind of displacement supposed .
Take , for instance , ( 3 ) as an example .
It is seen that , and the small separations are .
Thus if is really composite it should equal , and its value should be , it is already correct within very small error limits .
If , however , it is independent , the link 5647 , which is just as far as it can be from or , would be spurious , 23966 and 27783 would be linked to 20776 and 33431 to 38038 , and the error in the long link would be divided between observation errors on 20776 and 38038 , say and , giving respectively and .
All this argument , however , goes on the assumption that only one displacement can take place at each step .
Thus in passing from 27783 to 33431 the existing may displace to and link be , a line less than that observed .
The next link would be when 38038 is corrected by 2 and then the transformation to is completed .
That these double displacements take place is.not unlikely since the successive lines in our lists may be quite independent of the neighbouring lines , the chain links merely serving as a sort of Ariadne 's thread to discover them .
A corresponding treatment for all the sets gives : 3 .
.14 5 .
05 6 . .
19 8 . . . . . . . . . . . . . .
In the above the value of is the correction to be applied to the last line to make the long link or exact supposing the whole error thrown on that line alone .
In only are there indications that the small links show displacements the same as those given to , but this cannot be the case for both sets as the two last links must be the same .
The discussion of the cause of the enlarged links must therefore be left open until more accurate measurements are attainable .
There are a large number of more or less parallel linkages which unfortunately it .372 .
W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
has not been found possible to ibit o the maps in such a way that they can be seen at a glance .
There are clear indications that these are due in many cases to displacements of the lown kind in one of the terms .
In the maps the differences of the wave-numbers from the nearest lines before and after a given one are indicated .
Although these are intended for a different purpose , they indicate such displacements by the frequent appearance of the same separations , and in many cases the same on both sides , e.g. , in Ag , the mesh with one angle at 37233 is one striking example .
Amongst such separations in Ag may be given as illustrations 52 , 43 , 37 , 28 , 16 , where on is on is on is on is on is 16 .
But also 42 is the double link , , and may indicate an unobserved line .
As corresponding examples from Au may be taken the chain from 18449 .
A very exact example of lines at equal distances on either side is given by 26842 .
The adjacent lines differ from it , and on either side .
Now a displacement of on produces .
They are therefore ordinary displacements on either side of 26842 .
complete discussion of the whole spectrum must take account of these possibilities .
The present paper only deals with the establishment of the truth of the linkages efFect .
The majority of the linkages observed are represented in sixteen maps .
This method is in fact the only way in which the general connection can be visualised and realised .
They are based on the differences as found and consequently must contain a considerable number of chance coincidences , although for the reason given above ( p. 363 ) the number will be less than that given by the probability formula .
In the present state of our knowledge there is little to decide where they occur .
In a few cases the existence of a coincidence may be evident and be traced .
For example , the lines 15923 , 19113 , in the linkage which differ by , or apparently by a modified link , really belong to a doublet of the type , referred to in more detail later .
are respectively the and terms .
The satellite is 19738 , which gives the normal separation with 15923 .
They are in inverse order\mdash ; or their formula gives negative values .
The satellite separation and as it chances .
Thus ths apparent link is not realy so , it is a pure coincidence .
It is further possible that links attached to these lines may reproduce other examples of this pseudo link .
As a fact 15923 has a separation 17243 to and 19113 a link to 36367 , which suggests a spurious link between 33165 and 36367 .
In many cases it is easy to see that a spurious link must occur somewhere in a certain region .
For instance , if a linkage starting from one series line runs on into that of another by a certain sequence of links , one at least in that sequence must be a chance coincidence .
There are several cases of this , but astonishingly few when regard is had to the enormous number of links which exist .
One chief difficulty in settling the true sequence in a case like this is due to the existence of the link DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
37.3 modification just considered .
If only separations varying by one or two units from the calculated links are allowed , most of these difficulties , if not all , would disappear .
In the maps in certain cases a suggested false link is placed in square brackets .
The various cases are more itely treated in the notes attached to each map as they occur .
Another way in which the presence of spurious links is indicated is that of a complete cycle of links in the links do not each cancel out with another of the same kind .
Of true cycles , some examples have been given in Plate Another very long one is found in the map , ii .
and is considered on p. 388 .
Meshes also form a very numerous class of such small cycles .
Of false cycles a few examples may here.be cited for illustration from the general discussion below\mdash ; ( 1 ) In the linkage there occurs the cycle This could only be a true linkage if which cannot be .
The numerical values happen to be nearly equal in case of Ag only .
The link 2617 is not good , 3770-86 is practically exact and the 2460 belong to the long series(l ) on p. 366 .
It is probable , therefore , that or 2617 is the false link in this case , most probably 2617 .
( 2 ) In there is the mesh 963 .
962.04 apparently .
This cannot hold exactly , but it must necessarily be approximately true for any metal , since by their constitution must be near the mean of and .
Clearly in this case is the difference and is not a true link .
( 3 ) Another similar but longer cycle occurs in , be written In this case most of the links cancel and the set reduce to as in the last case .
The actual sequence can be followed on the map of .
DR. W. , HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
THE It is found that , starting from any link belonging to the ordinarily recognised series , a series of other lines can be linked up to them in succession .
These linkages are exhibited in a.diagrammatic form in the maps appended .
In these the wavenumbers increase towards the right .
They are arranged in vertical columri for easy reference , the small links stretch from one column to the next , the links from one to second from it , and the long link over three .
In this way the distance gives il general idea of the magnitude , and as a rule each column will contain wave-numbers of the same order of magnitude , although not always , especially when the links run forward and backward several times .
The links are represented by straight lines , with the actual observed separations placed over them .
Intensities are indicated by small figures over and to the left of each wave-number .
These maps contain all the links noted to be related .
In addition certain more abnormal ones , which other evidence , such as meshes or other regularity , render it advisable to put in evidence , are also shown .
Also in many cases abnormal links are shown when they end a chain .
In this case their presence can lead to no errors in succeeding lines .
From these maps lists ( pp. 391-410 ) are formed of some of the lines expressed as sums of multiples of the , by considering them in succession with their links .
These lists explain themselves .
The first column is an ordinal number for reference .
When a chain ends the wave-number is printed in italics .
In that case the next line will start a new chain branching from a previous line .
The numbers attached in brackets to its ordinal number refer either back to a previous line from which it starts or forwards to the branching off of a fresh chain .
Figures in brackets after a wavenumber give the small changes in it required to make the links correct , the last column of all headed giving the corresponding O-C of wave-length .
The numbers in heavy type between the wave-numbers give the exact observed separa- tions .
The numbers in the columns under , \amp ; c. , give the multiples of these quantities on which the wave-number depends , and when the modified values with or enter , they are entered as .
For instance under the column for , would in Ag , .
The colum1ls headed or refer to the variable part of the series line from which the linkage starts .
For example , in the linkage stands for .
I. Silver . . .
Wave-lengths observed by RANDALL in the ultra-red .
The separation is or 25 too large .
If is taken as correct , .
Any single link attached to increase these values lands us in a region which has not been observed .
If a linkage exists it is therefore necessary to try the sum of two links .
It is found that there are two * This portion of the text should be read in combination with the respective maps ( to 18 ) and formulae tables ( pp. 391-410 ) .
'Ann . .
Phys 33 , p. 742 ; 'Astro .
Jour XXXIV .
, p. 10 .
DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SEIiIES .
lines , shown on the map , which arelinked to by and another by clearly , an link is quite a common starting one from and series lines .
Moreover the next link is just much displaced one way from as is in the other from , for .
Thus the line is exactly above .
In fact we have a case of " " series inequality The link - to the next is excessive , but it is entered because ( see list Nos. 3 , 4 , 5 ) the displacement in ( 3 ) is destroyed in ( 4 ) by adding a to it , thus making it normal , and then another on the ( 4 ) and on the link reproduces ( 5 ) .
So far all seems satisfactory but considerations come in , in connection with the line 18744 .
There is an analogous set of doublets of the type ( see also p. 372 ) in each of Au , Ag and Cu with the proper doublet separation for each element .
That they belong to the type with negative values in Au and Cu is proved by their Zeeman effect , but the pattern for Ag has not been determined .
The Au set show a satellite while Cu does not , as is usual with the first element in a group .
In Ag the doublet set are with respective intensities of 4 and 1 .
There is a line 18756 also of intensity 4 which naturally sesms to be , but its satellite difference is not sufficiently near a multiple of to give confidence .
The line 18744 , however , which has just been suggested as belonging to the linkage with such good evidence , might just as well bs taken for were it not that its intensity 2 is less than that of instead of equal to it or greater .
the other hand it makes the satellite difference exactly and so affords some evidence of its being .
There is also this in its favour , that both it and 18206 form a kind of combination with to give two lines in the ultra-red observed by RANDALL , whose wave-numbers are and .
How close this relation is can be seen from the followin At first sight RANDALL'S ultra-red lines look like Ritz combinations but it is not so , * since they should be written *All RANDALL'S other ultra-red lines are , however , RITZ combinations , though his allocation would seem to be .
For instance , using the notation employed throughout this series of papers , hc gives , , if so the sequence must be identical with the , as is clearly seen from the wiIlg :\mdash ; in which is affected with a possible error and 7965 of .
The equality of the and is pointed out by him .
The lines considered in the text ( given by him as and , but as the lines are not known it is not clear how he has ) tained their values .
If , however , 18744 is really in the ( 2)linkage , the second of the above would only be a coincidence , and would be a sum of links of which no other example is known .
The existence of the two combinations together forms a strong argument that 18744 is the line , and that the linkage relation is a coincidence . . .
In the map of this linkage there must be at least one spurious link shown by a complete false cycle , and two possible ones shown by two chains , one leading to and the other to .
The links from the mesh back to 17874 ( No. 6 ) are all good except the two successive and .
They are admitted , however , because they are an exalnple of series inequality deviating by the same amount in opposite directions from exact normality .
They clearly point to a displaced link for the first with the displacement annulled in the second .
With the long links stretching from to 28968 we get the enlarged values which seem frequently to be associated with the chief lines .
From 28968 two chains diverge , one through 31429 ultimately leading to through 30393 .
There are no clear indications to show where the false link occurs .
The first link is so large that at first sight it might seem to bs spurious ; but it falls in well with the others as may be seen from the formula list .
It would look as if the series of long links should be , i.e. , , , but that the last for some reason took at first but completed the extra in the next link .
The formulae in this chain , as will be seen , involve very few displaced links beyond those it started with , due to the above indubitably enlarged links .
In fact such only enter in No. 21 and in 25 , 26 , which latter form part of a mesh .
On the other hand , as will be seen below , the formulae based on show very good links up to 29514 and even beyond .
Provisionally I have placed the spurious link as that between 29514 and 30394 with in place of to 28968 and considering the other branch chain ffom it , the second link is small , but is naturally accounted for by the long e\mdash ; which entered as due to 3779 after ( 3)\mdash ; changing to the same value as the second link 3776 which entered immediately after it .
The links and ) appear at first sight too abnormal , they are exactly explained by a return of the two long links after , which became equal in No. ) to the value of that before .
From .30806 starts the false cycle referred to above .
In order to keep the map as open as possible the line appears in two places in it\mdash ; here and at the top of the same column .
In the absence of sufficient knowledge of the laws governing links it seems impossible to point out where the pseudo link occurs .
The long links 3774 , 3778 which usually appear successively here appear separated , although a set appear as if a part of a missing 3778 .
This consideration would therefore favour DR. W. .
HICKS : STUDY OF SPECTRAL .377 see as .
Again the succession oflinks of would appear to signify a real succession .
There would then remain the two ) make the latter correct 29358 should be less .
This would and nearer their normal values , but would make too large .
On the other hand ) is about a displacement on ovisionally , therefore , firlse link is 29,358 and 31815 .
The eyidence for this location is not , but it is f:trengthelted somewhat from the fornlulae .
Tbese ( No. 41 ) that the ; introduced after are now cut out by others , pointing to their being parallel sequences .
Furthel the set of modified links from 3080 produce in the end a change or in hich oll modification bas disappeared .
Nos. 41 , , and 43 afl*ord an example of series inequality .
The )( links attached as tags to the lines here considered fall naturally into the formulae .
given Nos. 4 and 49 depend directly on the fornlulae the lines to they are attached .
They , take the simpler forms rive if displacements are allowed in other terms .
The ) ( No. 3 is of , that it might possibly be the .
line of a er , in nalogy w the other lines .
But the line falls in naturally in its posit i connection with is possibly accidental ( but see below under ( No. 37 ) a side chain starts which apparently leads into the linkage .
As will be seen below when the linkages are considered , the links up to 33179 seem good .
would seem a good line possibly one of the links near it be spurious , ) or .
The difficulty must at present ) left open . .
The only apparent link from these is that refel'red to in the foregoing , ) from .
is uot a good observatiou ( possible error ) ) .
If it be corrected from , which is much better ( 9 ) by adding , th link becomes .
which is , and gives .
It is thus so close to the usual link which starts from the series lines as to it very probable that it is a real link .
If so , other lines from those allocated above to will probably be linked up , another spurious link will exist amongst the set .
From a link 377676 , goes to 30722 , which in the linkage next to related to to it is connected by a link Now is 92036 or an exact link , the exactness of to uncertainties of the measurements , nnlst be pure .
If ) ) the difference would be , still within limits of .
is thon formed mesh :\mdash ; 919.58 3776 .
oxvrI . .
W. M. HICKS : CRITICAL ) ) OF SPECTRAL SERIES .
The conclusion to be drawn is that 26945 ] been wrongly allocated to by RUNGF It is further in favour of this that the difficulty as to the limits for the and series to in [ Part III .
p. 404 ] is got rid of .
link 9 ] ) cones in in the cycle from 30806 in which a number of doubtful links occur .
This one is clearly real , and it suggests that the adjoining one on map for is a false one .
lines 27140 , 28103 would then come into the system .
map for has been drawn thus : ] .
The linkage from is far the largest , over 2801ines appearing as linked together .
In many regions kage is so complicated that it seems impossible to ibit it in one map .
maps are used to show the connections , a chain leaving a one line and passing on to the same line on another map .
Amongst such a large number there must be many coincidences , and in fact the presence of some of is by cycles in which the links do not identically annul one another .
also few cases of ins appear to pass on to other independent linkageH .
It will be necessary , therefore , at first to Jnake c attempt to clear up these ambiguities .
The following cycles occur:\mdash ; 1 . .
In map there is a chain from 37574 ( col .
k ) to 43720 ( col .
) with li , :\mdash ; ( ii .
) col .
to connected by the If these are supposed all normal , their sum would be in and in ( ii .
) .
Now } .
Although not an absolute proof , this with total links in ( ii .
) is almost convincing that the pseudo link lies in the chain in .
But the links in are most of them good modified going back 43720 ( see map ) .
The 54 is partly due to error in the previous and a further on the line 37282 , which makes the better .
is or is large , but if 3350 is less ( all the other links converging on it become normal , viz. , link ) is ? .
The next link is larger than the majority .
It is .
We might be doubtful about this were it not that the next is and their sum or two normal links ( example of last is .
They all , therefore , apparently fit in displacement in links is , however , so small that little weight an be to their agreement .
Now a glance at the probability curves ( Plate 4 ) shows that 9 occur with frequencies less even than the calculated chance DR. W. .
HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
probability .
The spurious link is therefore probably one of these .
But if the link is retained there is a sequence through.37574 of and the regularity of which would seem to have a , and would include 921 as real .
Provisionally , therefore , or ( or both ) taken as spurious .
2 .
38577 in ( ii .
j ) to same line in .
here are two cycles in this case .
In ono the complete cycle of links is , startiug from ( ii .
) , ( 43720 ( ii .
) pass to ) .
In this the second is .
It is so excessive that it should scarcely been included .
We shall prob bly be right in rejecting it .
other is , starting from ( ii .
) , ( pass to ( iii .
) ) ( pass to ) the link .
may be marked as spurious .
3 .
The cancelling of the link 966 in the also allluls two other cycles , one a long one from 35435 in ( ii .
g ) to , to ( iii .
) , and other from .
through 33506 .
4 .
29737 appears both in and ( iii .
) .
The cycle is annulled by olnitting the large link ( ii .
) after 35150 .
The linkage connected , however , with 29737 probably belongs to to be considered immediately .
5 .
29084 on ( ii .
) is connected across to link , thelatter line belonging to 29737 just considered .
The cycle is annulled by lnitting t abnorlnal link after 35435 .
6 .
There is also the case given as an example on ( p. 373 ) and there considered .
With the links indicated above left out , the linkage is now singly connected , except in the case of true cycles .
There remains , however , one chain which leads on to the linkage .
The line ) .
in the cycle ( 5 ) just considered , is itself a part of the system .
It is linked to by a mesh , and it is practically certain that this connection is real .
In this case a certain set of lines attributed to the linkage must really belong to and a link must occur .
29084 is joined to 3170.3 by ) .
It is impossible to notice the sequences of this link with the links on the left hand top cornel of ( iii .
) without being convinced that they form a connected cluster .
It is difficult , however , to see where the false link comes anywhere on the chain up to which is connected in several ways with lines , or at least with lines attributed to .
Further discussion is deferred until the linkage is considered .
With the lines in the map now singly connected it will be possible to write down their formulae in terms of the values .
As , however , in such a large number of lines there must still remain a number of spurious links , and as the existence of one of these will invalidate all the formulae for lines occurring later , only the first sets starting from and are give here .
From a single chain stretches which is remarkable for the long series of enlarged links .
They are all close to their mean , which only differs from by 33 , and at the same time the link to is .
This chain , therefore , appears to support the theory of link .
W. M. HICKS : OF AL SERIES .
displacements adopted for the purpose of indicating their modified values .
The formuLe given in the list agree very ] with the observed values and depend , as is seen , on displacements of and displacements in and .
In the list our displacement convention has been departed from in No. 13 and dependent lines by assuming a displacement in .
This makes the link 3777 cut out a former link and reduces the formula to two terms .
is as far as it can be from an link displacement , and the suggested formula only gives a error .
From 28600 a long link connects to the whole succeeding lines through 31221 .
Some doubt might be felt as to its reality , but it is accompanied by an enlarged link , which suggests a parallel inequality .
The formula for 31221 gives .
It may be simplified .
to , giving possibly within observation limits .
But ifso , the error is such that it entirely upsets the parallel inequalities in the succeeding mesh .
It is , therefore , not adopted , although the line 30214 has the analogous form with .
Further , this error makes the parallel link into , i. e. , would replace ?
by one as far as possible from a displacement value .
The double mesh from 31221 is noticeable for the great modification in the links .
They are , however , examples of parallel inequality which demand with each link some displacement continued for each case .
In the table it is supposed to be on the l ) link .
It is possibly , however , on the term .
From 30415 the list passes from to and is continued to the strong line 34446 .
The linkages .
I have not been able to find any links attached to or .
They exist , however , for and . .
giving satellite difference , but the measures are not very accurate , especially that of .
There is a considerable linkage in which no cycles appear .
It contains , however , the line 29084 , referred to under the Plinkage , as connecting with lines in it .
It is itself a clear line , as the linked mesh in the map shows .
In order to test the links into the map ] ines , the formulae have been calculated up to the strong line 38234 ( see ( iii .
) ) .
The modifications of many links would seem to point to displacements producing about .
For instance in the first eleven links there are three examples , , the links below Nos. 1 , 4 , and 9 .
It is possible that they occur in the sequence and not in the links themselves .
For this reason and because the actual lines may be considerably in error , displacements have not been introduced into the formula , and errors up to or 2 have been admitted , whilst changes from the satellite sequence to the chief line sequence have been allowed .
But as a fact does not seem to be much demand for these displacements either in or , and the linkages are distinguished from those for the and linkages by their absence .
In No. 2 there is an example of a link acting as a pure displacement so that the three lines , , exist .
The link after No. 6 has DR. W. M. HICKS : A CRITICAL STUDY OF TRAL SERIES .
been taken as normal , although its amount is o11 account of the large errors required in 5 , 6 caused by neglect of the apparent displacement in the fourth link .
The link between Nos. 18 , 19 be the usual modified , but the arrangement adopted to from the satellite to the main would seem preferable .
A change from to also comes in Nos. 24 , 27 .
It is curious to note that all the chains terminate with changes near the end to , with the exception of those ending at 32062 , 35044 , and the doubtfully linked 32936 .
From No. 38 the chain which runs into the through some link .
The formulae are written down as far as the strong line 38234 , and so all the fall in very naturally and give small values .
The false link is probably , therefore , beyond 38234 .
This last 1nay be written wholly without displacement , those in disappearing in it .
The list is then continued to embrace the other lines to this chain .
It is interesting to see how the links fall in naturally , with the possible exception of No. 50 , which may be a coincidence . .
The only line obseryed is , but with the large possible error of .
It differs , ] vever , from the value as calculated from the series formula by .
If it be taken as about correct with a small satellite difference values of and would be about .
From approximate values it will be seen that linkages extend as shown in the map .
The table of formulae for lines are practically all normal links with small errors .
It will be noticed that No. 7 continues the ssion of , by case of 29514 , No. 3 , as connected with the linkage , has been there considered , and it was suggested that the false link was at .
The table gives the formulae for a few lines placed in the linkage on the that it real link .
It is seen how simply the formulae come in , but it is only possible to the large error which had been attached to 29514 , and this again has been caused by the adopted value for .
in appear more complicated because of the abnormal links ; the changes in each table from line to line are of course the same .
II .
In the case of gold the series lines in the arc are few , and eve these nre not certain , with the possible exception of the two lines , Tlley are considered in [ III .
, p. 404 ] , and way in which the constants there determined reproduce the links found in the spark spectrum affords very strong confirmation of the methods there employed .
The methods are based on an identification of the , lines .
These , therefore , will be take1l for granted now .
, but with no feeling of , two lines for , .
Their position and calculated denominators show that they range well with those of Cu and Ag , but there is no further evidence beyond their doublet separation .
They , however , form points for long linkages which l be called by their names .
The gold linkages .382 DR. W. .
HICKS : A CRITICAL STUDY OF SPECTRAL are many of very long , and naturally the presence of mere coincidences with link is to be expected to a greater extent than in silver .
The result is that in many cases a chain will appear to run from one linkage to another .
No attempt , however , has been made to locate the spurious , as such identifications would be even more uncertain than in silver .
It will be sufficient to indicate where a chain runs on from lnap linkage to another . .
At the first glance the presence of a large umber of meshes involving the long links is noticeable .
In the notes to the maps it will be advisable to follow the lists so far as the formulae are written down .
The residual differences between the observed and calculated values might in many cases be reduced by admitting displacements , but it has been thought better in the present state of knowledge to seek for simplicity a first approximation than exactness where the observation errors are quite unknown .
The separation of is practically exact .
There seem to be several links about 3188 , which have been introduced on the evidence of meshes .
From 3199 is regarded as and the next link as , but the two together can be regarded as in the value for 22813 and succeeding lines .
Also the 3199 and the 3191 from form a good example of series inequality , their sum being practically an exact .
The successive links 11347 and 14931 No. 7 would not have been admitted singly , but the excess of the first and equal deficit of the second point to a displacement in 28525 which does not recur in the preceding and succeeding lines .
It may be noticed that the order of operations is to by a displacement in the limit , then a modified link , then by another displacement , again by a modified link .
In the list as originally drawn there was no further displacelnent which should be 5635 ahead .
But there is a line 28439 which is ahead .
This is just as much in deficit from a link as the previous 3188 from .
It is therefore a link with 3188 made 1lormal , and clearly comes naturally into the system .
This again has a modified link .
If the law continued an extra displacement in the limit would give a line ahead , but none is found .
It would correspond to term .
The sequence is also continued back to 23966 by a modified and thence by displacement to 20768 .
Thus from 20768 we have the sequence\mdash ; ( say ) , .
HICKS : ITICAL '383 As supporting the theory of links , mesh 26000 ( and No. 4 ) is interesting .
For ) link is easily explained the displacement in link its to the line .
In fivct the sum of the two links is the link is not eIltel.ed as it is so small , but there are other exarnples of this small ) in this and the ' appears to be lnal associate with ) an nlesl .
Starting from fonr separations close to 1 or .
link is too much out to be troduced o its own account , but it forms part of lnesh whose corresponding link is evidently .
It is taken as displacement SOlnewhel'e in 42644 I not makes long link correct , but with the ance o on ) .
at the same time makes the separation to 6698 a correct link .
alternative ( No. 15 ) is on the supposition that is produced in the by ) Perhaps should stop at ils the next link ) , like the one eferred to aboye , is too mall .
But the next link is too small for by about the same alnount , and the hay t ppe rance parallel may possibly be due to the lJat ttions are due to ences b one line by and ) by nd HASCHEK .
If this be granted the next comes right with with error Again starting there is a collection of meshes , chiefly of .
They illustrate the variations from normal links chiefly in .
A able peculiarity is that in most of these meshes a line is linked by ) to one corner , in cases ) to 4 below its nol'mal value .
The regular presence of one of links in each vould seem to exclude ally explanation as a chaltce .
But they further exemplify the theory of displaced links in the defect from the normal values of is not due to themselves but to a concon mitant of a displaced or to a former value .
Thus in to its value line 39404 apparently linked to 45043 by eally takes the link , the in 45043 are annulled .
Also wipes out the same is above 39404 .
The line 2130 is of intensity 8 .
the rule the strongest lines are connected with link is satisfied there should be a mesh with outside observed limits .
The lin es 011 this mesh ( Nos. )5 to 28 ) appear also in the map .
From :31921 we get series of links similar displacements 2491 No. adl the previous nents itnnulled .
In 39404 above there is similar mple of toppil the displacements cut out .
( c.3 ) the sses to a separate to escape crowding .
In the next two lines we displace in ( links ) llornlality in the .
The linel from is affected with consideruble possible observation error .
, continued DR. W. CRITICAL STUDy from this , but with increasin doubt to spurious links .
In fact the long succession links in series in the lower part would seem to point to connection with a diffuse linkage .
presence of such a spurious link is shown directly also by the appeal.ance of the line ) in two places , cols .
and indicating a cycle . .
The linkage as shown in the llap is smaller than in case of Ag .
This is doubtless due to the fact that in both ca , the chains start with the long series of links .
Ag several of these in succession lead up to the ultra-violet region from which chains run back again towards longer wave-lengths .
In the case of Au , however , two of these long links carry the observed egion , and if the Au linkage is analogous to that of Ag , the points from which linkages would run back again are beyond the observed .
Possibly some may exist amongst HANDKE'S ultra-violet , but the observed possible rors i are so excessive , due to the very large multipIe of as well as to large possible values of , that it is best to neglect them .
There are several links in the neighbourhood of 17260 , where , and of .
It is possible these be links modified in some way not yet known .
In the lists , however , attelnpt has been made to explain all these modificatiollS by displacements on the links in general .
Starting , if the first link is without any modification , there must } ) an error in the line 20100 which is hardly credible .
If , however , and in the formula have displacements , the error in this line is reduced to .
The line 60270 is from HANDKE .
As the , it is just possible that the link from it to the previous line might be a normal or link .
The link from 54617 is about 10 in deficit and may be spurious , and in fact it leads on into the X linkage .
Tt is , however , , but this and those beyond have been excluded from the list .
In support of this further evidence is available from the consideration of the linkage below .
A mesh with clear links is based on and .
formulae seem to run naturally down to 18432 , which as in several cases ends the chain by all the displacements .
From 34669 chain leads to the lnap .
It old seem impossible at present to determine where the spurious link comes .
It may be noted that whilst 34669 contains there are lines on either side linked by and .
he curious double mesh attached to 40305 contains very abnormal links , with the calculated error which the ) gives to 40305 , the links from it form two pairs of parallel inequalities , and , which deviate equally in opposite directions froltl their itl values .
The lines are probably really connected .
T'he .
In [ III ] lines lotted to are with the satellite too faint to be observed , but a satellite start botl these lines .
addition there are the lines of } ) to ones in Cu and Ag already referred to ( p. 372 ) .
DR. W. M. HICKS : A ) OF SPECTRAL SERIES .
They are The Zeeman patterns of two been observed by HARTMANN .
His measures , though not perhaps very exact , seem to fit in with the type , the Zeeman patterns of the first two being represented in the notation suggested by by and respectively .
These are the patterns for and , so that the set are in the reverse order to the usual one , and the relative intensities agree with this .
This means that calculated as first lines in a series they have negative values as is the case when the first line is calculated from the formula .
If the limit be taken to be the same as for the diffuse , in agreement with its normal separation of , the denominator of the becomes with a mantissa about below that of .
The satellite separation of is produced by a satellite difference in the denominators of 27 .
Starting from these , a linkage of very great length extends , requiring five maps to represent them .
But it also contains the and lines , has connections with other linkages as well as possesses a certain number of false cycles .
There must therefore be many spurious links .
No attempt is made to locate them fully as was done in the case of silver , partly because they are more numerous and also because the greater difference between modified and normal links renders the entrance of spurious values more easy and their detection more uncertain .
The whole linkage is given in a set of five maps in which the line at which a chain leaves one map or appears at another is indicated .
As the origin of the set is not certain it is represented by the letter X. The satellite difference of the three lines forming the starting point is .
Now there are a large number of lines with a difference of about 600 which look as if related to this in the same way as the separations depend on the satellite .
It is too small to be in analogy with the general case of series in elements of other groups , but a similar diminution is apparent in Cu and Ag .
Many of the lines showing this are connected to links in a peculiar way .
In several cases lines differing by 600 have a line at being weaker than and looking like a satellite .
The difference between and is 14337 .
Now there are a very large number of of lines with this difference , indicating apparently the existence of fainter satellites the not observed .
Where such a difference occurs with a line in any of the maps it is indicated by a short broken line\mdash ; to the right when added and to the left when deducted .
As one example of the difference 600 , 'Phil .
Mag XXXI .
( 1916 ) , p. 171 .
VOL. CCXVII.\mdash ; A. 3 DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
a map is given starting from 18069 which is behind 18668 in Au X. , and is also attached to a typical link .
No formulae , however , from this map have been discussed .
In the map X. .
appears a congery of meshes , two of which seem to afford some support for the suggestion as to the connection between 600 and 625 .
The meshes in question may represented as follows:\mdash ; 14934 .
14935.17 The value 14911 would never have have been admitted were it not for the evidence of the numerous ) meshes in this region .
Their values may be explained as follows .
The line 33595 depends on ths sequence .
The line 33619 is greater , and this is the value of a displacement on the term .
If remains unchanged , and the lines depending are thus displaced , the satellite separation becomes less , i.e. , about 601 .
In this case the 14911 are not modified links , but are normal links with a simultaneous displacement on the or .
The type separation of 600 is produced by the new and the unchanged A very common difference found in the Au spectrum is 1000 .
Where it enters an arrow is to the number representing the line , to indicate its existence .
If bs the first diffuss , the limit of one of the series .
It is noticeable that .
The 1000 is therefore a link based on the diffuse sequences , the consideration of which has been excluded from the present paper .
There are also a very large number of separations equal to that of the satellites , viz. , 82 .
It will be instructive to write down the formulae for short regions of the X maps , but no useful purpose would be served by attempting to do this throughout .
Starting from the ) line 15923 in .
ii .
, we find the long series of links already referred to ( No. 4 on p. 366 ) .
It is noticeable that the last three sum to or three exact links .
At 38457 the chain passes to .
iii .
, and 38457 is a starting point for the very remarkable congeries of meshes already referred to Plate .
The formulae for only one of these is written down , but all the links cluster round 17257 , and it looks as if they are sums of single displaced links .
The link from 21201 ( No. 8 ) has only been admitted on account of the mesh .
I also comes naturally from the formula for 21201 by annulling displacements , , in the 3rd and 4th columns of the formulae tables , to .
At 38457 also a chain links on to by a link .
This link , however , is one of three DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL in eries , the last in a mesh .
They are probably real .
The next link , , is clearly a true link .
Of the two next to the second is a very close , and it would appear , therefore , that the spurious lin is , which is about as far as possible from a modified link .
There is also a long chain stretchin from 21558 in map ii .
to , which will be considered later .
The linkage is now taken up from 21558 with a chain of a large number of links requiring few modifications .
Nos. 14 , 15 , 16 , are parts of two meshes , equivalent to parallel sets of three lines at a distance 31977 .
The second mesh is the case taken as an example on p. 367 to illustrate the reality of displacements .
The opposite sides of the mesh are exactly equal , so that it is highly improbable that errors of observation enter to any extent .
The values of the four links are as far as possible from lnodified ones , and the displacements must arise in some other way , the lines 41325 , 41320 being displaced in opposite directions from .
The formula is given for this and the two displacements entered under .
There is an analogous mesh at the bottom of col .
attached to 25813 , but two different links .
The sides whilst .
This region is a maze of links which have a look of reality , even modified links constantly repeat themselves in similar arrangements , for instance 11345 , 14935 , 17250 .
Ths formulae are continued from 38127 by a chain through them from which the formulae for the other lines can be easily written down .
The chain from 18985 in the annexe is parallel to one from 18577 .
From the line , 19113 , only a short chain extends with a problematical link at the end .
It depends of course on instead of At No. 25 the satellite is taken up .
At 17410 on the diagram the chain passes from .
3 to for convenience of .
The part from 17410 to 41866 appears also in the map , and the formulae for L7410 , 28755 are Nos. 20 , 21 .
In this chain it should be noted that a line\mdash ; an ellhanced one\mdash ; is collnected to by a link .
The Zeeman pattern , as found from HARTMANN'S measures , is viz. , that of a .
It is co1mected by the links to line for the typical pattern is 5/ 10 .
relatiorlship is certainly not to be expected , and throws some doubt on the link .
Formulae are given as far as 18985 .
The representation of the slightly abnormal links looks artificial .
The abnormality is clearly real , as they so frequently appear and in similar positions .
It is possible they are due to displacements in the VD with a kind of satellite displacement , but if so they are too small to be accounted for by displacements .
From 19738 another chain starts\mdash ; drawn in the lower part of the map\mdash ; containing again a number of the same abnormal links .
Nos. 25 , 26 , which are separated by a very exact link , have the equal abnormal links attached symmetrically , viz. , to No. 25 and to No. 26 .
The symmetry is striking , but the first leads to a connection with the linkage and the second is that on which the Zeeman effect thrown doubt .
The formulae for this chain have not been written down , partly beca.use of the prevalence of the small modified ) links , and also because a few false cycles show that some spurious links occur .
In the X. .
map we have an example of a true cycle from 23354 , . .
The linkages appear amongst the X maps .
The line appears in the X. ii .
map in col . .
From this a chain proceeds to the top of the map consisting , like other linkages , largely of links .
They fit in well and the formulae are simple .
Another chain extends downwards into lines which at first sight appear analogous to those in X. .
There is , however , a remarkable long cycle which rather seems to indicate that the majority of the lines in X. ii .
belong to the series .
The lin 28620 appears at the top of the map at and also at the bottom , shown in the inset map at .
A true cycle runs between .
Starting from , and supposing all the links normal , they run in the following order to in which each kin oflink is annulled .
Although not an absolute proof it produces conviction that all the links are real .
The whole series forms , so to say , the backbone of the whole map in which only the side out-liers may be doubtful .
Assumin this it is easy to suggest the position of certain spurious links .
For instance:\mdash ; The chain from the true line ( 15923 ) joins this cycle at the line 42016 .
7 by the link .
This joins the corner of a mesh which is undoubtedly an set .
This link is probably spurious and is marked doubtful .
It differs as much as possible from a displaced link .
( b ) From 26303 a connection passes to the map .
It is the case already referred to in the discussion of that linkage .
The connection must be spurious , and the false link is probably the 11332 on which doubt was there cast .
( c ) A long cycle also stretches from 24273 to , but this is a false one .
It only has one link in common with the other , , that connecting 42397 and to .
appears in .
iii . .
From it a chain runs down to 38457 which has been taken to be on an X chain .
The links , however , all seem good from the present point of view .
If the and X lines are correctly allocated there must be a spurious link here .
If the links are real then and X are not independent , and one at least has been incorrectly allocated .
It is at least interesting to notice that the formula for 38457 depending on involves no displacements and reproduces the observed value by .
From 26494 another chain is drawn upwards .
The formulae are written down for a chain up to a terminal at 26566 from which side lines can easily be written down .
The two long links are naturally represented by with simultaneous displacements in other links .
At 41500 there is a typical DR. .
M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
parallel inequality correcting both the links attached to it .
In this chain it is to be noted that the several ?
links are all normal with the exception of that betwee1l Nos. 2 and which requires a displacement on a link .
The , X. .
map contains a large number of symmetries showing meshes involving the 1000 and 14333 separations , or the diffuse links indicated by arrows and broken lines\mdash ; see , e.g. , the neighbourhood of and .
A very striking illustration of such symmetry is afforded by the system attached to 27758 .
It is represented in the following scheme:\mdash ; 14936.44 997 .
14936.25 The agreement in the values of each separation is very close and would be wholly so if 28755 were about 5 larger .
As the X. iii .
map contains the linkage the term VD occurs in the formula for 27758 and the 1002 is due to a displacement in this .
We have in fact two parallel linkages depending on and respectively .
A similar case is seen in the neighbourhood of X. iv .
close equality of the differences between 26842 and the next lines , and , is noticeable and has been already referred to .
Now is caused by a displacement on and by on the same .
As the VD enters in the formula as a negative quantity , the smaller wave-number would be due to and the larger .
The maps bring out certain general peculiarities quite clearly .
Amongst these may be mentioned:\mdash ; ( 1 ) The majority of the strong lines are connected to the link either directly or by an intermediary link .
More especially so in the case of Ag .
( 2 ) The curious frequency with which , in a mesh involving the link , a or link is attached at one angle .
In many cases we find or ) links attached to an link as if the whole of a mesh not been observed .
( 3 ) The fact that where a large number of links converge on a line , that line is in general of small intensity\mdash ; suggesting that the number of confi urations present emitting the frequency is small , because it is only slightly stable and splits up into others related to it by the link relation .
( 4 ) The large number of links of normal value entering in the linkages .
DR. W. M. HICKS : A CRITICAL STUDy OF SPECTRAL SERIES .
The evidence given in the preceding pages would seem to establish conclusively the existence of these links and linkages in the spark spectra at least of Ag and Au .
Of 740 lines in Au only 41 have no , 1000 , or 14833 links attached to them , whilst 100 others with links have not been allocated to the linkages exhibited in the maps .
It is noticeable also that of these 100 , the majority show the 1000 and 14333links , indicating that they have relations to diffuse sequences\mdash ; not here discussed .
Of 600 lines in Ag only 92 show none of the or links , of which 0-4 occur in the first 60 and last 30 , where the fainter lines which might complete links are not observed through photographic and instrumental difficulties .
In addition , about 110 are not allocated to definite linkages in the maps .
We should expect many of these\mdash ; from analogy with Au\mdash ; to have reference mainly to linkages related to diffuse sequences .
The complete problem of the relations involved and criteria for true and spurious links , is one involving a much more exhaustive treatment than that given in the present communication .
It will necessitate the consideration of the sequence links , and especially of lateral displacements , the existence of which the present discussion has clearly indicated .
The cause of the variations in the magnitude of the links themselves must be discovered and will require a more accurate determination of wavelengths and of possible observation errors .
These questions are reserved .
DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
391 DR. W. M. HICKS : CRITICAL STUDY OF SPECTRAL SERIES .
co co co co 01 \mdash ; DR. W. M. HICKS : A CRITICAL ) OF SPECTRAL RIES . .
CCXVII.\mdash ; A. 3 DR. W. M. HICKS : A CRITICAL STUDy OF SPECTRAL SERIES .
5 5 \mdash ; DR. .W .
M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
o$ .
W. M. HICKS : A CRITICAL STUDy OF CTRAL SERIES .
5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 ? ?
co co DR. .
M. HICKS : A STUDY OF SERIES . .
' .
' .
' DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
\mdash ; DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
99 5 * DR. W. .
HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
5 5 on ; 1c DR. W. M. HICKS : A CRITICAL STUDy OF SPECTRAL SERIES . .
5 5 5 5 ri DR. W. .
HICKS : CRITICAL OF 5 5 5 \mdash ; I I I I I I I I I I I I I -J 5 ' LIEO V : SlIOIH M DR. .
HICKS : A CRITICAL ) OF TRAL S } 5 .
W. .
tTICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
5 5 oo : DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
5 .
\mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; - co \mdash ; DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
DR. W. M. HICKS : A CRITICAL STUDY OF SPECTRAL SERIES .
on 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 ic Phil. , vol. 217 , Plate 4 .
: : : : . . . .
: .
: .
2 : : So / 4 : .
' . . .
Phil. Trans. , , vol. 217 , Plate 5 .
:. .
: . . .
: .
: :::. .
, 40 : . .
NOTES To ( a ) 43720 in the linkuge , two meshes on liuk .
showing congeries of meshes ; also see .
showing complete cycles of links .
A similar one 19988 in ( h ) In shows parnllelism and symmetry .
( i ) 21696 in ( ii .
) ; 31967 in , all links diverging from or converging on .
( j ) ,30722 showing seqnence ) ) .
31494 , 29216 in shows trallelism .
( to one line makes all four links normal .
tfick Phil. , A. El sS Phil. , vol. 217 , Hicks . .
Trans. , , vol. 217 , VOL. CCXYII .
Hicks .
S Phil. Trans. , , vol. 217 , Plate 9 .
Hicks .
Phil. Trans. , , vol. Phil. Trans. , , vol. 217 , Plcrte 11 .
Hicks .
Phil. Trans. , , Plate 12 .
Hicks .
Phil. , vol. 217 , Plate 13 .
a Hicks .
Phil , Trans. , , Plate 14 .
ALLX Hicks .
Phil. Trans. , A. , vol. 217 , Plate 15 .
Hicks .
Phil. , A. , , Plate 16 .
Ill Au .
VOL. CCXVII.\mdash ; A. Ph.il .
Trans. , , vol. 217 , Plate 17 .
Au .
a a a a a fIicks .
Ph.il . .
17 , a Au .
1

