It is easy to map an arithmetic expression onto a stack-based virtual machine and then evaluate it. Many interpreters, specifically Forth, are built on stack-based virtual machines. To speed up execution, some stack-based designs have been implemented in hardware [51]. Microprocessors have been built for the direct execution of Forth [38, ] and are being designed for Java [52]. A large body of theoretical work on how to efficiently execute stack-based programs on register-based CPUs [37, , , , ] and on optimizing code for stack machines [50, , ] exists. The R-code interpreter will be stack-based and the techniques for optimizing stack-based programs and to speed up stack-based interpreters will be applied.