Michael Somos has been developing polynomial-valued versions of the theta and sigma functions. These are somewhat related to Somos sequences, but more general. His work so far is at http://somos.crg4.com/wxyz.gp It's more than I can grasp in one sitting, but it looks promising. Rich