The paper presents way for generating of a structural code, the basis for treating of the method for recording and reading of data in the storage, which can decrease the size of memory in one and a half time and, consequently, increase its indexes of reliability, dimension and capacity of the energy in few times. The synthesis code securely corrects modular errors and can be applied for the control of workability of the units of memory. Also we can see here fault-tolerant structures of the memory for 8-, 16and 32-positional formats of the data notation.