Some new results concerning tubular sets are presented, with applications to the convergence of the Polya algorithm in the contexts of simultaneous approximation and approximation of multivariate functions by univariate functions. (The Polya algorithm constructs a best uniform approximation from the limit, as p ->• co, of best L p approximations.) In 1913, George Polya proposed an algorithm to calculate best uniform approximations to continuous functions by polynomials [8]. This algorithm utilizes the continuum of L p spaces, and is known as the Polya algorithm. Current usage of the Polya algorithm is primarily theoretical. In [2] it was used to show that best uniform approximations exist, in [6] its limit was shown to be a Lipschitz uniform selection operator, and in [3] and [8] its convergence was studied in new contexts. The theme underlying these studies is the evolution of L p to L x as p -> oo. In [7], a sufficient condition for the convergence of the discrete Polya algorithm was presented. The present note continues the discussion of this condition and presents applications to the theory of multivariate approximation. In particular, it generalizes the simultaneous-Polya-algorithm-convergence result of [12], and it establishes the convergence of the Polya algorithm in a general context which includes the approximation of multivariate by univariate functions [1].