4 Mr. C. W. Merrifield .
Sums of the Series of the [ Nov. 17 , rence of these bands amongst tliG Fraunhofer lines , to warrant the publication of this note .
We have not yet succeeded in determining precisely the conditions under which the reversal can be produced at will .
The most complete reversals of these bands were obtained by the use of the arc of a Siemens ' machine , in a crucible of magnesia , fed with a considerable quantity of cyanide of titanium .
The photographs in this case show a very complete reversal of the five bands near L , and of the two strong bands near N , and a less complete reversal of the six bands , beginning at about wave-length 4215 .
No other metallic cyanides have given , when introduced into the crucible , any such reversal ; nor does a stream of cyanogen led in through a perforated carbon produce the effect .
Various other nitrogenous compounds have been tried , but the only one which has given us anything like the effect of the titanic cyanide is borate of ammonia .
Some photographs taken immediately after the introduction of borate of ammonia show distinctly the reversal of the group of bands near L. In one case when metallic magnesium had been put into the crucible , the photograph shows a reversal of only that part of the series which is nearest to the magnesium group , indicating that the reversal is due to the bright background supplied by the expanded magnesium lines .
There can be little doubt that the greater stability of titanic cyanide and boron nitride than of other nitrogenous compounds , has some influence upon the result ; and the difficulty in producing the reversal at will is in securing an absorbent stratum of sufficiently high temperature and at the same time a sufficiently luminous background .
The circumstances which secure the former condition almost always produce in the arc a still more intense radiation of just those rays which are absorbed , without that expansion of the lines which shows out the absorption in the case of so many metallic spectra .
The photographs are , however , conclusive evidence that it is possible to secure both conditions .
III .
" The Sums of the Series of the Reciprocals of the Prime Numbers and of their Powers .
" By G. W. Merrifield , F.R.S. Received August 8 , 1881 .
Euler has shown* that it is possible to sum the series of reciprocals of powers of the prime numbers , and he has calculated the values of these sums for the even powers .
I thought it of some interest to calculate the sums for the odd powers , and to evaluate a peculiar constant ( somewhat analogous to the Eulerian constant , \#151 ; 7=0-57721 56649 01532 86060 65 ) * " Introd .
in Anal. Infin .
, " vol. i , cap. xv , " De seriebus ex evolutione faetorum ortis , " pp. 221-252 .
1881-1 Reciprocals of the Prime Numbers and of their Powers .
5 which presents itself , in the series of simple reciprocals of primes , as the difference between the sum of the series and the double logarithmic infinity to the Napierian base e. The summation of these series was shown by Euler to depend upon the Napierian logarithms of the sums of the reciprocals of the powers of the natural numbers .
Euler 's reason for taking even powers only was doubtless that these latter summations could be expressed in terms of Bernoulli 's numbers , and of the powers of w , and could therefore be computed directly , without any actual summation .
A table given by Legendre , * and reprinted by De Morgan , t however , contains the whole series of the sums of the reciprocals of the powers of the natural numbers , even as well as odd , and I have made use of this table as the basis of my own work , being satisfied with the verifications of Legendre .
The first step was to find the Napierian logarithms of all the numbers given in Legendre 's table .
These were then combined , so as to give the summations of primes , by means of a theorem of Mobius^ used by Mr. J. W. L. Glaisher , S in calculating a corrected table of Euler 's values .
Instead of giving my own results only , I have thought it would be convenient to bring together the complete set || of\#151 ; 1 .
Legendre 's table of the sums of the reciprocals of the powers of the natural numbers , to 16 figures .
2 .
The Napierian logarithms of the numbers in that table , to 15 figures .
3 .
The whole series of the sums of the reciprocals of the powers of primes , to 15 figures .
I think it will be convenient also to give a very short note of the means of obtaining the results , as the best explanation which they can receive , and as saving some troublesome references to books not accessible to everybody .
Since the reciprocal of 1\#151 ; \#151 ; is the geometrical progression X lA+VV . . .
\gt ; X X* ar # * " Traite des Eonctions Elliptiques , " vol. ii , p. 432 .
f See his " Diff. and Int. Calc .
, " p. 554 , % Crelle 's " Journal , " vol. ix , p. 105 .
S See the " Transactions " of the " Association Eran\lt ; jaise pour l'Avancement des Sciences , " Havre , 1877 .
( I See Mr. J. W. L. GKLaisher 's paper in the " London Mathematical Society 's Proceedings , " vol. iv ( for 1872 ) , pp. 48-56 , " On the Constants that occur in certain Summations by Bernoulli 's Series , " for the summations of some allied series , namely\#151 ; 1*"* \#151 ; 2"w + 3~* \#151 ; 4rn -f . . .
( natural numbers ) , and 1~"* + 3~"n + 5"n'+ 7"* + . . .
( odd numbers ) .
{ \gt ; Mr. C. W. Merrifield .
Sum of the Series of the [ Nov. 17 , by giving * every prime value , and multiplying all the fractions on one side of the equations into one another , and all the progressions on the other side into one another , we have\#151 ; \#166 ; = l+^ + 5 + 5 + ^ + ^+ . . .
( natural numbers ) .
For the multiplication of the progressions brings together once , and once only , all the factors which make up each term of the harmonic series .
Moreover , this is true , not only of the simple numbers , but of nny powers , so that generally\#151 ; ( 1 ^ n ) 1(1 3 !
( 1 \#151 ; 5~w)_1 . . .
( primes ) \#151 ; 1 + 2-w4-3 'm + 4~"+ . . .
( natural numbers ) .
If n has any positive integral value except unity , the second side of this equation is always finite .
When 1 , it may be transformed ( as is well known ) into\#151 ; loge x-fy +\#151 ; \#151 ; \#151 ; V -f-\#151 ; ^\#151 ; _ ... .
\amp ; 2x 12^ 120*4 Calling the series of the reciprocal nth powers of the natural numbers S n , and of the primes 2 " , we have , upon taking the logarithms , \#151 ; ^Sw=2w-)--|22W-|-3'2gM-|- ; |24"-|- ... .
This would enable us to obtain the value of IS , if we knew the series 2 .
The theorem of Mobius , mentioned above , enables us to effect the reversal of this , and to obtain 2 in terms of IS .
This theorem , which is easily established inductively by indeterminate coefficients , is as follows : let\#151 ; .
~F(x)=f(x ) + hf(.x2)+kf(xS ) + lf(x4 !
) + ... .
then f(x)=F(\#187 ; ) -PX*2 ) -PV ) -*F(a\#171 ; ) + P(*6 ) - p(*7 ) +tp(*10 ) the law of the last series being that every term whose index contains a square factor disappears , the others being positive or negative accordingly as the index contains an even or odd number of prime factors .
Thus F(*4 ) , F(*8 ) , F(*9 ) , F(*13 ) all take the coefficient zero , while F(*6 ) and F(*10 ) are positive ; but all the prime terms , and such terms as F(*30 ) , F(*42 ) , \amp ; c. , are negative .
This at once gives\#151 ; \#166 ; 2n=ZS)i -|7Sgra\#151 ; iZS4re + \#163 ; ZS6"\#151 ; 1 \#151 ; ... .
1881 .
] Reciprocals of the Prime Numbers and of their Powers .
7 This is the formula given by Mr. Glaisher , and used both by him and by myself in the calculation of \#151 ; \#171 ; .
In the particular case of n\#151 ; 1 , we have S^log ( log x ) \#151 ; 031571 84520 73890 [ *=oc ] * I have not been able to identify this constant 0'3157 ... .
with any function of a known constant .
It does not seem to have any immediate connexion with the Eulerian constant \lt ; y ; for loge7=-0-5496 , 73=0-33316 .
There appears no reason why it should be commensurable with any simple function of 7 .
The following results may be useful for the purpose of this or analogous comparisons:\#151 ; 6y=1 -781072 , 6-y=0 -5614595 , e-2i=l-371241 , + e2i=0 -7292647 , + 6(1+20=1 -982347 , + g-a+sp\#151 ; o -5044525.+ It is worth while to remark that the S series I+2+3+4+ ... ... .
begins with unity , whereas the prime series 2 + 3+i+ .\#166 ; ... begins with and omits the unit .
The last two tables are given to 15 decimal places , having been calculated to 16 places .
Still , the last figure is not reliable .
The tables are not continued to very high values of n , because , for such values\#151 ; S , -l=log(S\#187 ; )=2\#187 ; :=2(S\#187 ; _1\#151 ; 1 ) \#151 ; 2 log ( S\#187 ; _i ) = 22\#171 ; _1 , true to 15 decimal places , or more .
* There may be some doubt as to the correctness of this expression .
This turns on whether the assumption which it involves is justifiable , namely\#151 ; log { S ( a ; )'-1 } - log ( log x ) = 0 when x = x , and consequently when both , terms on the first side of the equation represent divergent series .
This , however , is not a question which much affects the arithmetic .
+ Omitting the term log ( log 00 ) .
8 Mr. C. W. Merrifield .
Sums of the Series of the [ Nov. 17 , Table of the Sums of the Powers of the Reciprocals of the Natural Numbers , S('o)_re , to 16 Decimal Figures , from Legendre .
n. S(oo)-\#187 ; .
1 flog ( cc ) + ... ... ... . .
10-57721 56649 01532 9 2 ... ... ... .
1 -64498 40668 48226 4 3 ... ... ... 1 -20205 69031 59594 3 4 ... ... ... 1 -08232 32337 11138 2 5 ... ... ... 1 -03692 77551 43370 0 6 101734 30619 84449 1 7 ... ... ... 1-00834 92773 81922 7 8 1 -00407 73561 97944 3 9 1 -00200 83928 26082 2 10 1 -00099 45751 27818 0 11 1-00049 41886 04119 4 12 1 -00024 60865 53308 0 13 1 -00012 27133 47578 5 14 1 -00006 12481 35058 7 15 1 -00003 05882 36307 0 16 1 -00001 52822 59408 6 17 1 -00000 76371 97637 9 18 1 -00000 38172 93265 0 19 1 -00000 19082 12716 6 20 1 -00000 09539 62033 9 21 1 -00000 04769 32986 8 22 1 -00000 02384 50502 7 23 1-00000 01192 19926 0 24 1 -00000 00596 08189 1 25 1 -00000 00298 03503 5 26 1 -00000 00149 01554 8 27 1 -00000 00074 50711 8 28 1 -00000 00037 25334 0 29 1 -00000 00018 62659 7 30 1 -00000 00009 31327 4 31 1 '00000 00004 65662 9 32 1-00000 00002 32831 2 33 1 -00000 00001 16415 5 34 1 -00000 00000 58207 7 35 1 -00000 00000 29103 8 1881 .
] Reciprocals of the Prime Numbers and of their 9 Table of the Napierian Logarithms of the Sums of the Powers of the Reciprocals of the Natural Numbers , logeS(oc )-* , to 15 Decimal Figures .
n. logeS(oc ) \#187 ; .
1 ... ... . .
log ( log oo ) 2 ... ... . .
0 -49770 03024 70745 3 ' ... ... .
0-18403 41753 91491 4 ... ... . .
0 -07910 98730 67336 5 ... ... . .
0 -03626 22596 49228 6 ... ... . .
0-01719 43876 02658 7 ... ... . .
0*00831 46149 69275 8 ... ... . .
0 -00406 90663 07413 9 ... ... . .
0 -00200 63787 01528 10 ... ... . .
0 -00099 40808 65669 11 ... ... . .
0 " 00049 40665 33147 12 ... ... . .
0 " 00024 60562 78979 13 ... ... . .
0-00012 27058 18911 14 ... ... . .
0 -00006 12462 59468 15 ... ... . .
0-00003 05877 68496 16 ... ... . .
0 -00001 52821 42636 17 ... ... . .
0-00000 76371 68475 18 ... ... . .
0-00000 38172 '85979 19 ... ... . .
0 -00000 19082 10896 20 ... ... . .
0 -00000 09539 61579 21 ... ... . .
0 -00000 04769 32873 22 ... ... . .
0 -00000 02384 50474 23 ... ... . .
0-00000 01192 19919 24 ... ... . .
0 -00000 00596 08187 25 ... ... . .
0 -00000 00298 03503 26 ... ... . .
0 -00000 00149 01555 27 ... ... . .
0 -00000 00074 50712 28 ... ... . .
0 -00000 00037 25334 29 ... ... . .
0-00000 00018 62660 30 ... ... . .
0 -00000 00009 31327 31 . .
' ... . .
0-00000 00004 65663 32 ... ... . .
0-00000 00002 32831 33 ... ... . .
0 -00000 00001 16416 34 ... ... . .
0:00000 00000 58208 10 On the Reciprocals of the Prime Numbers , fyc .
[ Nov. 17 Table of the Sums of the Powers of the Reciprocals of the Prime Numbers 'Z(x)~H , x Prime , and takeu from 2 to oo ; to 15 Decimal Figures .
n. 2* ( .
1 r log ( log oo ) 1 -0 31571 84520 73890 2 ... + 0-45224 74200 41065 3 . . .
0T7096 26392 99444 4 . . .
0 -07699 31397 64247 5 . . .
0-03575 50164 83924 6 . . .
0-01707 00868 50637 7 ... 0 -00828 38328 56134 8 ... 0 -00406 14053 66518 9 ... 0 -00200 44675 74962 10 . . .
0 -00099 36035 74437 11 ... 0 -00049 39472 69104 12 . . .
0 -00024 60264 70035 13 . . .
0 -00012 26973 67528 14 . . .
0 -00006 12443 96725 15 . . .
0 -00003 05873 02823 16 ... 0 -ooooi 52820 26219 17 ... 0.00000 76371 39371 18 ... .
o -ooooo 38172 78703 19 ... o -ooooo 19082 09077 20 . . .
, o -ooooo 09539 61124 21 ... o -ooooo 04769 32759 22 . . .
, o -ooooo 02384 50446 23 ... .
o -ooooo 01192 19912 24 . . .
, o -ooooo 00596 08185 25 ... , o -ooooo 00298 03503 26 . . .
, o -ooooo 00149 01555 27 ... .
o -ooooo 00074 50712 28 ... .
o -ooooo 00037 25331 29 . . .
, o -ooooo 00018 62660 30 . . . .
o -ooooo 00009 31327 31 . . .
, o -ooooo 00004 65663 32 ... .
o -ooooo 00002 32831 33 ... , o -ooooo 00001 16416 34 . . . .
o -ooooo ooooo 58208 35 ... .
o -ooooo ooooo 29104

