The text is concerned with the definition and investigation of some classes of ordinal valued-functions; and with the determining of a particular assignment function O. This function assigns to each limit ordinal alpha smaller than a constant theta and belonging to the second number class a fundamental sequence [diagram] that is, a strictly increasing sequence satisfying [diagram]. The operations used to determine O are derived from a class of functions [diagram]. Since generalizes the standard ordinal arithmetic operations,some of the properties of the are studied, together with the operations gammaalpha which are generalizationsof transfinite sum and product. Also investigated is a related class of number-theoretic functions, in this context O is an arbitrary assignment bounded by o in place of [diagram]. It is proved that the functions are normal in the second argument, and these functions are compared with a hierarchy of normal functions obtained by Veblen's process of iteration. The provide a natural means of extending the notion of epsilon number, and some of the properties of the generalized epsilon numbers are presented. The function of [diagram] is normal, and the notation [diagram] is adopted for the sequence of countable fixed points of the function. The generalized epsilon numbers determine a hierarchical classification of limit numbers, and on this basis a normal form is determined for each, and thence the function O is defined by transfinite recursion.

The symbol map (of Goncharov) takes multiple polylogarithms to a tensor product space where calculations are easier, but where important differential and combinatorial properties of the multiple polylogarithm are retained. Finding linear combinations of multiple polylogarithms in the kernel of the symbol map is an effective way to attempt finding functional equations. We present and utilise methods for finding new linear combinations of multiple polylogarithms (and specifically harmonic polylogarithms) that lie in the kernel of the symbol map. During this process we introduce a new pictorial construction for calculating the symbol, namely the hook-arrow tree, which can be used to easier encode symbol calculations onto a computer. We also show how the hook-arrow tree can simplify symbol calculations where the depth of a multiple polylogarithm is lower than its weight and give explicit expressions for the symbol of depth 2 and 3 multiple polylogarithms of any weight. Using this we give the full symbol for I_{2,2,2}(x,y,z). Through similar methods we also give the full symbol of coloured multiple zeta values. We provide introductory material including the binary tree (of Goncharov) and the polygon dissection (of Gangl, Goncharov and Levin) methods of finding the symbol of a multiple polylogarithm, and give bijections between (adapted forms of) these methods and the hook-arrow tree.