12 .
" On certain Properties of the Arithmetical Series whose ultimate differences are constant .
" By Sir Frederick Pollock , Lord Chief-Baron of the Exchequer , F.R.S. &amp;c. This paper professes to investigate certain properties of the series of whole numbers whose ultimate differences are constant , and incidentally to treat of Fermat 's theorem of the polygonal numbers , and some other properties of numbers .
Its object is to show that the same ( or an analogous ) property which Fermat discovered in the polygonal numbers belongs to other series of the same order , also to all series of the first order , and probably to all series of all orders .
It also proposes to prove the first case of Fermat 's theorem ( that is of the triangular numbers ) from the second case of the squares ( which had not before been done ) , and to dispense with the elaborate proof of Legendre ( Theory des Nombres ) , finally , to prove all the cases by a method different from that either of Lagrange , Euler , or Legendre .
It is first shown that an analogous property belongs to all series of the first order ( viz. common arithmetical series ) .
The following propositions are then proved as the basis of future reasoning:1 .
Every triangular number greater than 6 is composed of 3 triangular numbers .
2 .
Every triangular number greater than 3 is composed of 4 triangular numbers .
3 .
Any triangular number may be expressed by the form ( a+ a+ b2 ) .
4 .
The sum of any two triangular numbers may be expressed by the same form .
5 .
Every number above 7 is the sum of four triangular numbers exactly .
6 .
Every number above 29 is the sum of three triangular numbers exactly .
7 .
Every multiple of 8 is composed of eight odd squares , and the sum of any 8 odd squares is a multiple of 8 .
8 .
The following general theorem is then proved : Ifp be any odd square , then Apz+BPY+Cp+ + v.Z p &amp;c. will equal 8 odd squares , if A+B-+CrD+ , &amp;c. equal 8 , or any multiple of 8 ( A , B , C , D , &amp;c. , z , y , x , &amp; c , may be any positive whole numbers ) .
9 .
It is a corollary from this , that in any system of notation having an odd square for its base , the sum of the digits will have the same number of odd squares as the number itself ; the number of odd squares being in each case the minimum .
10 .
Any number of the form 8n +4 is composed of 4 odd squares .
11 .
It follows from this that every number is composed of 4 triangular numbers , or 3 , 2 or 1 .
12 .
From this it is shown that every number is of the form ( a2+a4^-b2 , c2+C+d2 ) .
13 .
And that every number of the form 4n+2 ( if n be greater than 2 ) is composed of 4 square numbers , 2 even and 2 odd .
14 .
And that every number greater than 27 is composed of 8 squares exactly .
15 .
Every number ( beyond a certain small limit ) is composed of 2 triangular numbers +a square number , or 2 triangular numbers +a double triangular number .
16 .
A proof is then offered that in the equation 8n+4=4 odd squares , one of the four odd squares may be any odd square less than 8n+4 , and therefore 1 may be one of the 4 odd squares ; and if so , 8n + 4=1 +3 odd squares,. .
8n +3 =3 odd squares , from which Fermat 's theorem of the triangular numbers is an immediate corollary .
17 .
A proof ( by a tabular series ) is then suggested , that all the other cases of Fermat 's theorem may be deduced from the first , and that it is not necessary to use more than four terms greater than unity ( as discovered by M. Cauchy , see Suppl. to Legendre 's Theory des Nombres , p. 21 , 22 ) .
18 .
A general expression for a succession of series is then givenl , ( p+l ) ( 3p+ ) , ( l ) ( 6p+l ) &amp;c. ( p+ ) ; and it is proved that any number may be composed of not exceedingp+2 terms of the series , of which three only are required to be greater than unity .
If p=9 , the series is 1 , 10 , 28 , 55 , &amp;c. , that is every third triangular number beginning with 1 ; and every number of the form 9n+q consists of q triangular numbers not divisible by 3 ( if q be greater than 2 ) .
Ifp=8 , the series is 1 , 9 , 25 , 49 , &amp; e. 19 .
( The odd squares ) and every number may be composed , of not exceeding 10 odd squares .
Ifp=6 , the series becomes 1,7 , 19 , 37 &amp;c. ( the differences between the cubes ) .
From the continued addition of the terms of this series , the cube numbers may be formed .
Ifp=4 , the series is 1 , 5 , 13 , 25 &amp;c. , or 1 , ( 1 +4 ) , ( 4 +9 ) , ( 9+ 16 ) , &amp;c. The continued addition of the terms of this series forms the octohedral numbers , viz. 1 , 6,19 , 44 , &amp;c. Every term of the series is composed of ( p + I ) prior terms ; also if q be added to any term it will equal ( q+ 1 ) prior terms .
20 .
It is then proved that the property of the triangular numbers is not destroyed by adding any ( the same ) number to each terni , it is merely postponed , and commences at a higher number according to the magnitude of the number added .
21 .
The same is proved in respect of the addition of any common arithmetical series .
22 .
The paper concludes by suggesting a proof that every number may be composed of 4 triangular numbers , derived from the consideration that if the triangular numbers be indexed or numbered thus1 2345 &amp;c. indices .
136 10 15 &amp;c. An nos .
Any number between 2 triangular numbers can be formed by 4 triangular numbers , the sum of whose indices shall be not less than the sum of the indices of the 4 triangular numbers that compose the smaller triangular number , and not greater than the similar indices of the larger ; and generally ( after a limited number of terms ) the sum of the indices of any intermediate number will be exactly the sum of the indices of the smaller number : to illustrate this , all the numbers between 91 and 105 ( 2 triangular numbers ) are shown to consist of 4 triangular numbers , whose indices exactly equal 25 , which is the sum of the indices of the 4 triangular numbers into which 91 may be divided6 667 thus 91-=21 +21 +21 +28 the sum of the indices 6+6--6+7=25 ; and every number between 91 and 105 may be composed of 4 triangular numbers , whose indices added together will equal 25 ; but the nature of this investigation cannot be made intelligible without reference to the table itself , which the paper contains .
If this attempt is successful , the whole of Fermat 's theorem of the polygonal numbers may be proved without reference to Lagrange 's proof of the case of the squares ( the second case ) derived from the properties of the prime numbers .
The writer intimates an intention of making further communications on the same subject .

