@misc {IOPORT.03494450,
    author       = {Erd\H{o}s, Paul and Simonovits, Miklos and S\'os, Vera T.},
    title        = {Anti-Ramsey theorems.},
    howpublished = {Infinite finite Sets, Colloq. Honour Paul Erd\H{o}s, Keszthely 1973, Colloq. Math. Soc. Janos Bolyai 10, 633-643 (1975).},
    year         = {1975},
    abstract     = {[For the entire collection see Zbl 0293.00009.] Let $H$ be a fixed graph: $f(n;H)$ denotes the maximum number so that any edge coloring of $K^n$, the complete graph on $n$ vertices, which uses $f(n;H)$ or more distinct colors yields a subgraph isomorphic with $H$ each edge of which has a different color (a totally multicolored subgraph). The paper contains a reformulation of some earlier results, some new results, and some conjectures, all concerning the function $f(n,H)$. For example: Theorem 4 (a new result): Let $p \ge 4$. There exists an $n_p$ such that if $n>n_p$, then $f(n,K^p)= \text{ext}(n,K^{p-1})+1$, (where $\text{ext}(n,K^{p-1})$ is the maximum number of edges a graph on $n$ vertices can contain without containing $K^{p-1}$ as a subgraph). Further if $K^n$ is colored by $f(n,K^p)$ colors and contains no totally multicolored $K^p$, then its coloring is uniquely determined: one can divide the vertices of $K^n$ into $d$ classes $A_1, \ldots ,A_d$ so that each edge joining vertices from different $A_i$'s has its own color (that is, a color used only once) and each edge $(x,y)$ where $x$ and $y$ belong to the same $A_i$ has the same color independent from $x$, $y$ and $i$. Conjecture 1. Let $C^k$ denote the $k$-circuit. Then $f(n,C^k)-n((k-2)/2+1/(k-1))+0(1)$.},
    reviewer     = {J.E.Graver},
    identifier   = {03494450},
}