1: \begin{abstract}

2: A theorem is proved concerning approximation of analytic

3: functions by multivariate polynomials in the $s$-dimensional

4: hypercube.  The geometric convergence rate is determined not

5: by the usual notion of degree of a multivariate polynomial,

6: but by the {\em Euclidean degree,} defined in terms of the

7: 2-norm rather than the 1-norm of the exponent vector $\k$ of

8: a monomial $x_1^{k_1}\cdots \kern .8pt x_s^{k_s}$.

9: \end{abstract}

10: