In the present note certain known theorems on the latent roots of matrices are deduced from the fundamental theorem that a matrix A can be expressed in the form PQP-1, where P is non-singular and Q has zero elements everywhere to the left of the principal diagonal, and the latent roots of A in the diagonal. [The presence or absence of non-zero elements to the right of the diagonal is known to depend on the nature of the “elementary divisors” of the “characteristic determinant” of A, but in what follows these will not concern us.]

In connection with Bertrand's algebraical exercise of 1850, Muir remarks that it is not unlikely that the divisibility of a3 + b3 + c3 − 3abc by a + b + c had been previously noted, although. there is no record of the fact. Bertrand's exercise is to the effect that the circulant of the third order repeats under multiplication or, what is the same, admits of composition; the formulae of composition are stated in the exercise precisely as they would follow from Spottiswoode's theorem on the linear factors of a circulant. The whole of this is implied as an immediate special case, and indeed as one that any reader would construct at once, in an identity due to Lagrange, reproduced by Legendre.

The conception of the integral of one function with respect to another was introduced by Stieltjes in his classical memoir on continued fractions. He denned the integral as

and gave the formula for integration by parts.

Let R be a (1 — 1) relation between the members of two similar classes A, B1. It correlates the members of a subclass X of A to the members of a certain subclass Y of B1 and thus defines a relation ρ connecting X and Y. It is clear that ρ is a (1 – 1) relation and that it has the property (M). If X1ρ Y1, X2ρY2, then implies

In looking for a compact way of writing down the partial fraction formula in general, with repeated factors, I noticed how the expansion of a determinant by its top or bottom row suggested a method. The following gives a formula perfectly easy to write down in any given case where the factors of the denominator of the fraction are known. Incidentally it gives, as a determinant, the integral of a rational fraction f(x)/Q(x) where f(x) and Q(x) are polynomials, Q(x) having higher order.

Let the Nodal cubic have the equation

or,

Three points t1, t2, t3 form an apolar triad to the curve when they satisfy an equation of the form

The definition here used of the Stieltjes Integral is the same as that of a previous note, viz.:—

Let f (x), φ (x) be two real functions defined in (a, b) a finite interval on the axis of the real variable x. Let Δ1, Δ2, …, Δn, be a finite set of sub-intervals which together make up (a, b). Δrφ denotes the increment of φ (x) in Δr. Let ξr be any point of Δr, and form the sum

The result of dividing the alternant |aαbβcγ…| by the simplest alternant |a0b1c2…| (the difference-product of a, b, c, …) is known to be a symmetric function expressible in two distinct ways, (1) as a determinant having for elements the elementary symmetric functions C, of a, b, c, …, (2) as a determinant having for elements the complete homogeneous symmetric functions Hr. For example

The formation of the (historically earlier) H-determinant is evident. The suffixes in the first row are the indices of the alternant; those of the other rows decrease by unit steps. This result is due to Jacobi.

Introduction. In the present paper a formula will be obtained to express a Ferrers' Associated Legendre Function of any integral degree and order as a sum of a finite number of Associated Legendre Functions of an order reduced by an even number. When the order is reduced by unity, an infinite series of the functions of reduced order is required. Thus a Ferrers' function can be expressed as the sum of a finite or infinite number of zonal harmonics according as the order of the function is even or odd.

§1. The Lamé Functions of degree n (where n is a positive integer) may be defined as those solutions of the equation

which are polynomials in the elliptic functions sn x, cn x, dn x of real modulus K. Such solutions only exist for certain particular values of the constant a; there are 2n + 1 such values and 2n + 1 corresponding Lamé functions.

§ 1 Introduction. The formulae

where n, q, m are positive integers, are well known. I am not aware, however, of any paper dealing with the corresponding definite integrals of products of two Ferrers' Associated Legendre Functions of different integral orders and degrees. With the help of the formulae given in my previous paper in this volume it is possible to discuss these integrals. I have found in the case of the first integral that if the orders of the two associated functions are both odd or both even and if, in each case, the degrees are both odd or both even or equal, the integral has usually a non-zero value. If, however, the orders are both odd or both even and if the degree of the function having the greater order is less than that of the other, or if the degrees are one odd and one even, the integral vanishes. Further, if one of the orders be odd and the other even, and the corresponding degrees are also even and odd (or odd and even), the integral has a non-zero value; in all other cases it vanishes. The values of the second integral can be obtained in a similar manner. An example of the methods employed and the results obtained is given in the following section.

Turnbull : On differentiating a matrix, p. 126, line 12-

Professor W. P. Milne has shown that if a pencil of plane cubic curves cut in two triads of points which are apolar to the members of the pencil, then the other three points of intersection also form an apolar triad to the pencil. I propose to show how to obtain a simple geometrical construction for the third apolar triad though the cubics in this case are not perfectly general. The method of approach is by means of Grassmann's construction for a cubic curve and the use of apolar theorems established for a curve described in this manner.

The standard method of establishing rigidly the tests for a maximum or minimum value of a function of two independent variables, depending as it does on the use of Taylor's Theorem and on a very critical consideration of the Remainder in that theorem, presents difficulties so considerable that it is not surprising that most text-books on the Calculus frankly decline to enter on the discussion, and assume the necessity and sufficiency of the well known Lagrange's Condition. It is the object of this paper to show that a satisfactory proof of the tests may be given, from the purely geometrical standpoint, without recourse to Taylor's Theorem. The method requires only an elementary knowledge of the process of changing the independent variables in partial derivatives, and may therefore be introduced comparatively early in the Calculus course.