I. " On Newton 's Rule for the Discovery of Imaginary Roots of Equations .
" By J. J. SYVLESTER , F.R.S. Received May 4 , 1865 .
In the first part of my " Trilogy of Algebraical Researches , " 9 printed in the Philosophical Transactions , will be found a proof of Newton 's Rule for the discoyery of imaginary roots carried as far as equations of the 5th degree inclusive .
The method , however , therein employed offered no prospect of success as applied to equations of the higher degrees .
I take this opportunity , therefore , of announcing that I have recently hit upon a more refined and subtle method and idea , by means of which the demonstration has been already extended to the 6th degree , and which lends itself with equal readiness to equations of all degrees .
Ere long I trust to be able to lay before the Society a complete and universal proof of this rule -so long the wonder and opprobrium of algebraists .
For the present I content myself with stating that the new method consists essentially , first , in the discerption of the question as applied to an equation of any specified degree into distinct cases , corresponding to the various combinations of signs that can be attached to the coefficients ; secondly , in the application of the fecund principle of variation of constants , laid down in the third part of my 'Trilogy , ' and , in particular , of the theorem that if a rational function of a variable undergoes a continuous variation flowing in one direction through any prescribed channel , then at the moment when it is on the point of losing real roots , not only must it possess two equal roots ( a fact familiar to mathematicians as the light of day ) , but also its second differential , and the variation , when for the variable is substituted the value of such equal roots , must assume the same algebraical sign* .
By aid of the processes afforded by this principle , which admits of an infinite variety of modes of application , according to the form imparted to the channel of variation , and constitutes in effect for the examination of algebraical forms an instrument of analysis as powerful as the microscope for objects of natural history , or the blowpipe for those of chemical research , the problem in view is resolved with a surprising degree of simplicity ; so much so that , as far as I have hitherto proceeded with the inquiry , the computations , algebraical and arithmetical , which I have had occasion to employ may be contained within the compass of a single line .
The new method , moreover , enjoys the prerogative of yielding a proof of the theorem in the complete form in which it came from the hands of its author ( but which has been totally lost sight of by all writers , without exception , who have subsequently handled the question ) , viz. in combination with , and as supplemental to , the Rule of Descartes .
On my mind the internal evidence is now forcible that Newton was in possession of a proof of this theorem ( a point which he has left in doubt and which has often.been called into question ) , and that , by singular good fortune , whilst I have been enabled to unriddle the secret which has baffled the efforts of mathematicians to discover during the last two centuries , I have struck into the very path which Newton himself followed to arrive at his conclusions .
Received May 18th , 1865 .
Since the above note was sent in to the Society , I have completed the demonstration for the 7th degree , and in the course of the inquiry have had occasion to consider the conditions to be satisfied in order that a rational function of x , with r equal roots a , may undergo no loss of real roots for any assigned variation imparted to the function : for the theory of the 7th degree the case of three equal roots has to be considered , and the conditions in question are that the variation itself may contain the equal root a , and that its first differential coefficient may have the contrary sign to that of the third differential coefficient of the function which it varies when a is substituted for x-a theorem which is , of course , capable of extension to the case of an equation passing through a phase of any number of equal roots* .

