A Memoir on Integral Functions .
\#171 ; A Memoir on Integral Functions .
" By E. W. Barnes , M.A. , Fellow of Trinity College , Cambridge .
Communicated by Professor A. R. Forsyth , Se .
D. , F.R.S. Received July 25 , 1901 .
( Abstract .
) The memoir deals with the asymptotic expansion of functions with a single essential singularity at infinity in the neighbourhood of that singularity .
The term " integral function " is used as a translation of the French expression " fonction entiere .
" Part I opens with an introduction , in which it is pointed out that for each of the few integral functions whose detailed properties have been investigated , there always exists near infinity an asymptotic expansion valid in those parts of the region near infinity which are not at a finite distance from zeros of the function .
It is then suggested that wide classes of integral functions admit such expansions ; and it is pointed out that , if such a theorem can be proved and the expansions obtained , we may solve many questions relating to the " genre " of a function , to the nature of its zeros , and to the character of its coefficients when expanded as a Taylor 's series .
We may , in fact , to a large extent classify such functions by their behaviour at infinity .
After the introduction , a short account of the historical development of the enquiry is given ; and then the memoir proper commences with a formal arrangement of integral functions according to the laws of distribution- of the zeros as seen in expressions in Weierstrassian-product form .
When the 7ith zero an is such that it depends solely upon n and certain definite constants , and also such that the law of dependence is the same for all zeros , we call it a simple integral function with a single sequence of non-repeated zeros , or sometimes briefly a simple integral function .
The zeros are said to be algebraic , when they are given , when n is large , by a formula of the form r , ci c2 i " L nr* J where pis a rational positive quantity , and pi , p2 , . .
are rational positive quantities arranged in ascending order of magnitude .
If p 00 J 00 1 is such that 2 \#151 ; -r-r- converges , and 2 =1 la\#171 ; |p+e \#187 ; =l| \#174 ; n|p e diverges , however small the positive quantity e may be , the function is said to be of order p. We can form a scale of simple integral functions arranged according Mr. E. W. Barnes .
to the dependence of an upon n. When an is , for n large , to the first approximation a transcendental function of n , the integral function is said to be one of transcendental sequence .
A function with a finite number of simple sequences of zeros can be built up of a number of non-repeated functions each of simple sequence .
When the nth zero of a function of simple sequence is repeated a number of times dependent upon n , we call it a repeated integral function .
The typical zero may require more than a single number to define its .
position in the series of zeros to which it belongs ; the integral function is then called one of multiple sequence .
Finally , it is pointed out that the category to which a typical zero / belongs may be indefinitely complex , and that , therefore , we can expect to lay down no general law relating to all integral functions , which is not a disguised truism .
In the memoir we confine ourselves ; substantially to simple functions with repeated or non-repeated } sequences of zeros .
Part II contains the theory of divergent and asymptotic series , f Poincare 's arithmetic definition is first given ; the function J ( z ) admits 1 the asymptotic expansion ai a0 + \#151 ; + z + \#151 ; + in certain regions where \z\ is very large , if the sum of the first n terms be sn , and the expression tends to zero as z tends to infinity in those regions .
The difficulty of this theory is pointed out , and it is shown that a theory dependent on analytic functionality is more tractable and no less precise .
A series a0 + ctiz + . . .
+ + ' . ?
*\#171 ; of finite radius of convergence can , by an extension of a method due to Borel , be interpreted by means of a contour integral for all values of the variable outside the circle of convergence , except those which lie along the straight line joining the singularities to infinity , and proceeding directly away from the origin , provided all the singularities of the function represented by the series lie on such a line .
The integral is called the " sum " of the divergent series .
The process of summation is not unique , but must always lead to the same result , namely , to the function which is the analytic continuation of the function represented by the given series when convergent .
On the A Memoir on Integral Functions .
other hand , the series for those values of z for which it is divergent is not the arithmetically asymptotic expansion of the function .
At this stage it is natural to invent some extension of the ideas thus employed , and to apply that extension to the case in which the series has zero radius of convergence .
It is shown that this new problem is radically different from the old one , the number of solutions being infinite .
The solutions given by the process of integration are , however , functions which have zero for their essential singularity , and which admit the given series as an arithmetically asymptotic expansion in certain regions whose apex is that singularity .
^ * We classify divergent series according to their order ; a series is said nv/ dT to be of order k when -\#151 ; ~ is finite and not zero .
n16 Series of the first order can be " summed " by the same process as that employed for series of finite radius of convergence .
Series of higher order can be " summed " either by successive repetition of this process , or by means of auxiliary functions derived from the higher hypergeometric functions .
An application is made to the Maclaurin sum formula , and it is shown that this formula may be used to give an asymptotic expansion for 2 \lt ; f\gt ; ( n ) , when m is large , when \lt ; f\gt ; ( z ) is an integral function of of order greater than or equal to unity .
Finally , the rearrangement of asymptotic series is considered , and Part II closes with a theoretical account of the possibility and nature of the asymptotic expansion of an integral function near its essential singularity .
In Part III actual asymptotic expansions of wide classes of simple integral functions with non-repeated zeros are obtained .
The three standard functions employed are\#151 ; where p is greater than unity and not integral , and is an integer such that p + 1 \gt ; p \gt ; p ; Pp ( ?
) = n x [ i + ^j , where 1 j ( -\#171 ; )\gt ; np(z ) = i n where p is an integer ^\gt ; 1 .
^ -A Memoir Integral Functions .
It is shown that , asymptotically near z = oo , we have log Pp 0 ) -\#171 ; Up sin P l'g - -t log \#163 ; + 2 -\#151 ; ( ps)- .
8 = 1 IogQp ( ?
) and logRp ( y ) \#166 ; sin Trp 2p log 2ir \#151 ; ^ log z + 2 ' 8= \#151 ; p ( -)\gt ; -\gt ; F(s/ \#187 ; S2* ( *\gt ; log*+(-)p 1logger-i-log 2jt / \gt ; 2/ \gt ; ' ( -)*-1F(g/ p ) + 2 ' tTve a%umeTbm F " * 8ubstantia % functions of negaI t is then shown that these expansions can be generalised so as to apply to wide classes of functions , notably those with algebraic sequence of zeros .
General formulas are given which symbolically contain the expansion of all integral functions of the type considered , which are such that the Maclaurin sum formula can be applied to the function expressing the dependence of the nth zero upon And as expansion of'f * f""Cti ' " with transcendental sequence , the asymptotic .
v* .
a , \#187 ; = \gt ; L \#171 ; " J IS obtained comjDlefcely .
Part IV deals with the asymptotic expansion of repeated integral functions of simple sequence .
It is necessary to take an extended definition of " order " in the case of such functions , and then actual expansions are obtained for standard functions__ ( 1 .
) Of finite ( non-zero or zero ) order less than unity ; ( 2 .
) Of finite non-integral order greater than unity ; ( 3 .
) Of finite integral order greater than or equal to unity .
Subsequently symbolic formulae are given for all integral functions of the prescribed type ; and , as an example , the asymptotic expansion of a repeated function with a transcendental index is considered The formulae are verified in the case of the G-function .
I art V is devoted to applications of the previous expansions to the problems mentioned in the introduction .
A knowledge of the asymptotic expansion of a function serves to determine the number of roots which it possesses inside a circle of given huge ladius .
If the function is of order / \gt ; , the number of roots within a circle of large radius r is to a first approximation Io\#171 ; SZs On Areal Induction .
where 4\gt ; ( r ) is the maximum value of the function on the circle in question .
When p is integral , this approximation becomes log 0 ) ' log r A number of theorems are given relating to the order of functions combined from functions of given order .
It is shown that a function is of the same order as its derivative , and that from the knowledge of the zeros of a function we may derive an approximation to the large zeros of its derivative .
If a function is of " genre " p , its derivative can at most have only p zeros beyond those given by the extension of Bolle s theorem .
The memoir closes by suggesting other applications of the theory .
" On Areal Induction .
" By George J. Burch , M.A. , F.B.S. , Beading College , Beading .
Beceived September 10 , 1901 .
In my paper " On the Belation of Artificial Colour-blindness to Successive Contrast , " * I described a modification of some experiments by Sigmund Exner .
A disc , half black and half white , with a very narrow slit at the junction of the two halves , revolves in front of an incandescent lamp .
Another lamp throws a light upon the face of the disc .
W hile it is rotating slowly , the incandescent filament of the lamp , seen through the slit as it passes across , looks bright against the white card , but when a certain velocity is reached it appears as a black thread against a brighter background .
I suggested as the explanation of this phenomenon that the intense light of the filament induced an amount of fatigue out of all proportion to the sensation excited by it , so that the less fatiguing illumination of the white card produced a greater effect on the senses than the sum of the sensations due to the filament and the subsequent light .
In a recent paper , !
however , Dr. Shelford Bid\vrell has described an experiment having an important bearing on the subject .
By a suitable arrangement a disc of green light is projected on a screen ; this is followed by a somewhat larger disc of white light , and the cycle is completed by an interval of darkness .
The novelty of the experiment consists in the addition of a black spot in the centre of the white disc , so that when the apparatus is worked the site of this spot is illuminated five or six times a second by green light only , * ' Roy .
Soc. Proc. , ' vol. 66 , p. 212 .
f ' Roy .
Soc. Proc. , ' vol. 68 , p. 277 .

