]\gt ; 285 ] X. On J. J. .
L. , F.R.S. : Professor of of ) ' JAMES HAMMOND , Rcc } ived Junc 11 , \mdash ; Read Junc 16 , 188 INTRODUCTION .
IN the year 1786 SAMUEL , Professor the Uniyersity of Lund in Sweden , showed how by an extension of the method of it was possible to deprive the genel.al equation of the 5th degree of three of its terms without solving an equation higher than 3rd .
By a well-understood , however singular , fiction , this discovery was ascribed by to one of his own pupils , a certain SVEN SOMMELIUS , and embodied in a thesis humbly submitted to himself for approval by that ) ] , as a preliminary to his obtaining his degree of Doctor of Philosophy in the University The process ) effecting this reduction seems to have been overlooked or forgotten , and was subsequently re-discoveled many years ] ater by Mr. JERRARD .
In a report con tained in the ' 's of the British Association ' for 1836 , Sir showed Mr. RRARD was mistaken in supposing that the adequate to taking away more than three terms of the equation of the 5th degree , but upplemented this somewhat unnecessary refutation of a result , priori to be impossible , by an extremel .
yaluable discussion of a question raised by Mr. as to the number variables required in order that any system of equations of derrrees in those variables shall admit of being satisfied without solving any equation of a degree crher than the highest of the given degrees .
In the year the senior author of this memoir showed in a paper in KRONECKER'S ( better known as CRELLE'S ) 'Journal ' that the trinomial equation of " " Reduction of the Quintic Equation\ldquo ; was by ) Rcv .
} HARLEY , F.R.S. , in bhe ' Quartcrly Journal of Pure and Applied Mathematics , ' vol. 6 , 1864 , p. 45 .
full title of tlIc rhesis , as given by Mr. HARLEY ( see 'Quart .
Journ. Math pp. 44 , 45 ) is as follows : " " 13 .
cum T ) .
Meletemata quaedam matbemalica circa nsformationcm ationum a ] , quae conscnt .
Ampliss .
Facult .
Philos. in Rcgia Academia Carolina Praeside D. , Hist. Profess. Reg. \amp ; Ord .
publico Lxamini modest subjicit SVLN obltI \amp ; Lundensis .
Die Decemb. , ) , L.H.Q.S.\mdash ; Lundac , typis Berlingianis ative miuima ( minima , egard teight)here terms tmployed.iven tname ohese taken away from the equation occupy consecutive places in it , beginning with the second .
Mr. JAMES HAbfMoND has quite recently discovered by the method of generating functions a very simple formula of reduction , or scale of relation , whereby any one of these numbers may be expressed in terms of those that precede it : his investigation , which constitutes its most valuable portion , be found in the second section of this paper .
The principal results obtained by its senior author , consequential in great , measure to Mr. HAMMOND'S remarkable and unexpected discovery , refer to the proof of a theorem left undemonstrated in the memoir in ' CRELLE ' aboye referred to , and the establishment of certain other asymptotic laws to which HAMILTON'S Numbers and their differences are subject , by a mixed kind of reasoning , in the in apodictic , but in p art- also founded on observation .
It thus became necessary to calculate out the 10th Hamiltonian Number , which contains 43 places of figures .
The highest number calculated by ( the 6th ) was the ) umber 9 , which comes third in order after 5 ( the BRING Number ) , 11 and 47 being the two intervening numbers .
It is to be hoped that some one will be found willing to undertake the labour ( considerable , but not oyerwhelming ) of calculating some further numbers in the scale .
The theory has been " " a plant of slow growth The LUND Thesis of December , * For instance , an equation of not lower than the 905th may be transformed into anobher of that degree , in which the 2nd , , 4th , , 6th , , terms are all wanting , by means of the successive solution of a ramificatory system of equations , of no one of which the exceeds 6 , whereas by the Jerrard-Hamiltonian method this nsformation could not be effected for the general equabion of degree lower than the Hamiltonian Number , , 923 .
So for the analogous removal of 5 consecutive terms the inferior limit of degl'ee of the equation to be transformed would be by the one metbod , but 44 ( the lowest possible ) by the other .
In the case of 4 terms HAMILTON could not avoid being aware that 11 , the 4th number which I have named him , might be replaced by 10 , as the lowest possible inferior limit of the equation to be transformed .
In the 3rd section , communicated to the soCiety a the 1st and had to press , the empirical element is entirely eliminated , and the results educed to apodictic certainty .
ON HAIITLTON'S NUMBERS 1786 matter of a couple of Report of , with the tract of Mr. JERRARD therein referred to , and the memoir in 'CRELLE ' of December , constitute , as far as we are aware , the nplete b raphy of the up to the present date .
S1 .
On the Asymptotic of Vumbcrs of IILTON their Differences .
Consider the xble : . . . . . . .
Any line of figures , say , . .
, in the Table given , to form the subsequent line , . .
, we write If we call the term of the line , the general law of deduction may be expressed by the formula where means the coefficient of in ON HAMILTON'S ] The question arises as to whether it is possible to deduce the miltonian Differences , or to deduce the miltonian Numbers , directly in a continued from one another without the use of any intermediate numbers .
} HAMbIOND has shown that it is possible , and has the that it is the Numbers of Hamilton , and not the Numbers , which are subject to a very simple scale of relation .
These found , of .
course Differences become known .
This is contrary to what one would have expected .
riori , one would have anticipated that the determination of the Hypothenusl Numbcrs would have preceded that of their sums .
I leave Mr. HAMMOND to give his own account of his mode of obtaining the formula of reduction , which , by a modification , I find , be expressed as follows :Using E to denote the Hamiltonian Number mented by unity , so that , . .
; and to signify the coefficient of in ; then , for any value of greater than unity , . . .
tant part in the systematic treatment of the problem , first appcars in my memoir on the subject in he 100th volume of 'CRELLE .
' It is proper also to notice that what I call the Numbers of ( at all cvcnts thosc to the number 5 ) are nob the smalJest numbers requisite for illing the condition ave specified .
llcr numbers will serve to satisfy that condition taken alone ; but wllcn Huch smaller numl ) arc tituted for the resolving equations ] be less simple , inasmuch they will co1ltai]l a number of equations of the higher recs than when the largor Hamiltoniau numbers are ) This distinction will be found fully explained in memoir cited , and the smallest numbers for HAMILTON'S are there detcrmined for quatiollS of dogrecs extcnding to for all of up to 8 inclusive .
I have added nothing ( for there othing to be added ) to the formula of expressed by the equation where , supposing the letters , . . .
, to bc in number , means the number of letters required in order that it may be possible to satisfy , according to ocess employed by ( in conformity with a stipulation of JERRARD ) , a system of cquations of tions of degree equations of degree , . .
, equations of the degrce 1 , without solving any single equation of a degree higher than .
This formula , applied times succossivcly , will have the effcet of abolishing and causing to depend on , . .
, . . .
connccted with , . . .
by means of the formulae given at the commenceme1Jt of the prescnt paper , but instead of the letters , . . .
I have used the letters , . . .
It is presumable that the Hamiltonian numbers would be fonnd much Icss amcnable to algebraical treatment than the Hamiltoluan numbers proper ; for numerical equalibics and incqualitics have to be taken account of , in determining them , which have no placo in the detel'mination of the latter numbers .
IILTON , as already stated , oxpress ] alludes to the reduction of 11 to 10 , but with that exccption has avoided of finding the absolute lowcst numbcl of lcttel .
S required in that a system of equations ( expressed in terms of those letters ) of given rees may admit of satisfied without the necessity arising to solve any equation of a bighor degreo the highest of the given ones .
ON HAMILTON'S NUMBERS .
sum may be taken as a vertical sum of line-sums or as a horizontal sum of columnsums , and , although for licit values of each sum has a finite value , the two finite values are not identical , just as a double definite integral may undergo a change of value when the order of its integrations is reversed .
* I noticed at p. 478 of the 100th volume of ' CRELLE ' that the value of any Hamiltonian Diflerence divided by the square of the preceding one was always greater than , and stated as morally certain , but\ldquo ; awaiting exact proof that this ratio ultimately becomes .
By aid of Mr. HAMMOND'S formula for the numbers , I shall now be able to supply this proof , and at the same time to show that the ratio of a Number to the square of its antecedent ( which , of course , converges to the same asymptotic value ) is always less than that limit .
We must in the first place prove that in the series . . .
the absolute value of each term is greater than that of the one which follows it .
In this , I shall avail myself of the property of the Hypothenusal Numbers disclosed in the process of forming the triangle given at the outset of the memoir , viz. , that is greater than Let us suppose that the law to be established holds good for a certain value of For the sake of brevity , I denote , . .
by , . .
We have then \mdash ; 1 *Professor CAyLEy has brought under my notice a not altogether dissimilar , but perhaps less striking , phenomenon , pointed ont by CAUCHY , that , although the series is convergent , its square i. e. , is divergent .
The fortunate circumstance of the two ratios in question being always respectively less and greater than the common asymptotic value of each of them enables us to find the value of the constant in the expression , which is asymptotically equivalcnt to the half of the Hamiltonian or Number by a method exactly analogous to that of ustions for the Archimedian constant correct to any required number of decimal places .
See end of this section ( pp. 298 , 299 ) .
ON HAMILTON'S NUMBERS .
it will be true universally ; for in ) the succeeding series the term we have called will be higher than the term 6 in the scale 3 , 4 , 6 , 12 , 48 , .
Hence .
For the initial values of .
[ When represents any term beyond the first it is very easy to prove , but too tedious to set out the proof , that the sum of all the terms after the first in series equated to \mdash ; 1 will be less than ; so that , except in the case stated , For the series 12 , 48 , 924 , .
we have seen that Hence , for the series 48 , 924 , .
, or But Hence , and Q. Hence , when , are at an distance from the origin , Hence , also , ultimately which proves the theorem left over for " " exact proof\ldquo ; in the memoir referred to .
It is convenient to deal with the halves of the sharpe ) Numbers of HAMILTON , which may be called the reduced miltonian Numbers , and denoted by with a subscript , or , when required , by ) , , . .
halves We have then *Numbers increased by unity may conveniently be denominated sbarpeued numbers , and numbe ] diminished by unity flattened numbers .
ON NUMBEILS .
and we have preyiously shown that at all events when is not lower in the scale than The fraction arises from our having substituted for the inferior value but , the higher we advance in the scale , the nearer approaches to and is ultimately in a ratio of equality with it .
But , if we had written ) for the coefficient , which now stands at - , would have been\mdash ; .
In like manner , as and are traveUed on in the scale , and indefinitely near to and , i.e. , , so that the coefficient of in the superior limit approximates indefinitely near to , i.e. , and the two limits of which have been obtained become where ultimately and are infinitesimals .
* Hence it follows that the ultinuate value of i.e. , when Let , . .
represent the nalves of the Hypothenusal Numbers in the triangle given at the commencement of the paper , i.e. , the differences of the numbers which we have called , .
Since and , and Obviously , therefore , as a first approximation when , are very advanced terms in the hypothenuse , Let us write for a second approximation .
* As a matter of fact , it will be found that , as soon as and ttain the values 6 , 24 , may be taken as a supe , rior limit .
It may be noticed also , to prevent a wrong inference being drawn the above expressions , that , as will hercafter )pcar , is an of the order , when is infinite .
ON bhat led me to undertake the very considerable botlr of Hamiltonian Number in order to dednce from it the value of ) .
taken for , we may proceed to ascertain a further term in the value of expressed as a function of For , calling and we have neglecting .
12375 .
The value of and of , we may feel tolerably certain , from the Law of Squares , that ( will bo somewhere in the of the tenth part of 0035 , and is about1234 , so that the probable value of ( Sq ) is Thus we have found . . .
, the only moral doubt being as to the of closelless of propinquity of tlJe coefficient of to the decimal1234 .
For the be1lefit of those who wish to carry on the , I give the numerical results which have been elnployed in the arithnnetical tions : * It is reduced to cert in the s secticn .
The exact value of coefficient of , left blank in the text , is ( in scction 3 to ) c. , 1110 decimal CCLXXXV I. 2 ON HAIIILTON'S By means of this formula , for .
values of , and retaining so much of the two corresponding tion s of as is com1llo to botl ] , we can precisely to ) desired number of places of , as shown in the following Table , in which 18 and 24 are taken as the terms of place zero in the respective series .
, 438 , , 401 , , 463 , , 413 , 204348 , 41881398318 , 4088 , 4089863 204810 , 4090 , Hence , if we now change the , taking and 2 as the zero terms , we have approximately and which gives As a verification , since should lie between I8 and 24 ; and , as a matter of fact , a rough calculation gives .
, . .
, which is about midway between the two limits.\mdash ; J. J. S. S2.\mdash ; Proof of the the Snccessive Dete of each in turn of HAMILTON'S Nnmbers , om its Antecedents .
Let , .
, , , where the coefficients of the various powers of are the numbers set out in the triangular Table at commencement of this paper .
If , in general , we write the coefficients of , expressed in terms of those of , are as follows:\mdash ; * See Note 1 , p. 312 .
ESTE R AND MR. J. HAMNIOAT 1 .
2 .
3 Now , when multiplied by .
, gives Comparing this with , we see that the difference of the two expressions is which is equal to .
Thus Multiplying this equation by , where . . .
, .
31 ON HAbIILTON'S we obtain which gives , when we write successiyely , . .
in the place of ; ; .
Jfence , by addition of these equations , we find where it has been assumed that it is possible to assign to ( previously undefined ) such a value as will make the last of the above equations , identically true .
That this can be done is obvious ; for , if in that equation we write for , , and their values , , and then , on making , the equation becomes .
Thus the general value of is given by the equation .
is equivalent to here , , . .
being the Hypothenusal Numbers 1 , 2 , } } , . .
we hnve ON HAhIILTON'S NUMBL1 This equation is easily transformed into from which , as Professor has pointed out in this moir , by equating coefficients of all powers of from to , we can obtain the successive values of The general formula .
arises from equating the coefficients of .\mdash ; J. H. S3.Sequel to tloe Asymptotic Theory S1 .
The relation , etc. previously obtained supplies only the two first terms of the remarkable asymptotic development where is any assiglled integer and is of a lower order of ma-nitude than the lowest power of in the series which precedes it .
This may be easily established as follows:\mdash ; By the scale of relation proved in the section we have \mdash ; terms .
maximum order is that of Let , no . .
therefore . . .
and . . .
\mdash ; . . .
Therefore @ Note 3 , p. 312 .
July 28 , 1887 .
ON HAMILTON'S NUMBERS .
It is interesting to notice that the formula remains arithmetically true for finite values of and , proyided that is not less than 24 , when we each Hence , from what has been shown above , this equation we ma.y write .
( equations ) , where , . . .
are all of them finite ( and , as a matter of fact , of no consequence for our immediate object , positive proper fractions ) .
For , ultimately , ( see p. 307 ) , and consequently the finiteness of each is a direct inference the general principle previously applied in the case of the Applying this result to the equation previously given , it follows that . .
( where is finite ) where lies between 1 and 2 .
The series of negative powers of is iously less than , and the -series , which follows it , is less than the finite quantity 2 , i. e. , .
Hence ere is a number between fixed limits , and , the rank of , is of the same order of magnitude as Iog .
This equation contains as a consequence the asymptotic theol.em to be proved ; for , using to denote any positive integer , Hence , remembering that is of the same order of magnitude as ] , and that which is of alower order of magnitude than , it follows that for all.values of is ultimately in a ratio of eqnality with , which is the theorem to be We have thought it desirable to obtain the formula for its own sake , but , so far as regards the proof in question , that might be obtained more expeditiously from the expression given for without ) the series It is easy to tain the ultimate value to which converges .
In the first place , the series of fractions .
to terms ( where is the rank of q ) may be shown to be always finite , and consequently , when divided by , converges to zero .
For we know that . . .
Hence the last term of the series . . . .
Hence the finite series .
for a double a fortiori reason is less than the infinite geometrical series [ In fact , from S1 ( p. 299 ) it may easily be shown that the last term of the series , so that the sum is really less than ] Hence , retracing the steps by which has been oblained , and observing that differs from ) by a finite multiple of , we have ultimately \mdash ; .
If , then ( using to denote the half of the pened x Hamiltonian number ) , we write , and understand by the infinite series , it is easily seen that the principal part of , regarded as a function of and , is ON HAMILTON'S UMBERS .
or , since \mdash ; is a positive quantity , at all events when or It will be found also on trial that this formula remains true for all the values of inferior to 462 .
Thus Hence , universally , But we know that We msy therefore write , where is some quantity between and 1 .
Similarly , where , . .
are also positive fractions .
When and become infinite , Hence the ultimate value of is , . .
all of them conyerge to the value This agrees with the result previously demonstrated ( p. 295 ) , and is the ting point of all that follows .
We know that letters , . .
, being used to denote the halves of the augmented Hamiltonian Numbers , they are connected by the scale of relation this inequality been true ouly for values of sufficiently great , it would have been enough for the purposes of the text .
stands for the terms , .
PROFESSOR SYLVESTER,.MR .
J. HAMMOND Considering to be of the order we may reject the term , which is of zero order , and ; T. Hence , rejecting terms of order than ( which have , however , to be retained obtalnmg the subsequent approxlmations ) , 3 ; i. e. , when is infinite .
Again , writing for its expanded value , we have er ' : rejecting the terms , . .
in the expansion of because the of none of them is superior to zero .
We now write so that Hence Order 1 ' ON HAMILTON'S NUlIBHiRS .
Since ( ultimately ) , the terms of Order 1 ( which are the only ones with which have to do at present ) are ultimately equal to ; or , giving its ultimate value , to , or to the same order of approximation to .
Hence , ultimately , We use this result to obtain a closer approximation to than r , and to find the relation between the general values of and Thus , assuming we have , ultimately , .
Consequently , as becomes indefinitely great , converges to the value Now ultimately ; and similarly .
\mdash ; ultimately Hence , ultimately , We may therefore write ( where ultimately ) .
But and therefore Moreover whence it follows that ( where ultimately ) .
previously obtained by observation in S1 ( pp. 296 , 297 ) .
It will , of course , be understood that in the above and similar passages ths sign to be interpreted to mean " " is in a ratio of equality ON NUMBh'RS .
Here the terms of Order are ultimately equal to which , when and receive their ultimate values , and , becomes From this it follows immediately that ( rejecting terms of an order of magnitude inferior to that , ) The law of the indices in the complete development is easily deduced from the relation The terms carrying the arguments .
furnish the indices 2 , 1 , , 1 , , . .
, which , arranged in order of magnitude , become 2 , , . . .
Thus , calling and and respectively , the expansion for in terms of will be of the form where has all values from to , and does not exceed , i.e. , has all positive values from to or , according as is even or odd .
But , besides this expressed portion of the development of a Hypothenusal Number , say , as a function of its antecedent , , there will be another portion , consisting of terms with zero and negative indices of having functions of for their coefficients , which observation is incompetent to reveal , and with the nature of which we are at present unacquainted .
The study of HAMILNON'S Numbers , far from being exhausted , has , in leavin our hands , little more than reached its first stage , and it is believed will furnish a plentiful aftermath to those who may feel hereafter inclined to pursue to the end the thorny path we have here contented ourselves with indicating , which lies so remote from the beaten track of research , and offers an example and suggestion of infinite series ( as far as we are aware ) wholly unlike any which have engaged the attention of mathematicians .
J. J. .
and J. H. *Agreeing closely wibh what had been previously found by observation in S 1 ( p. 297 ) .

