The TI-83+/84+ calculators store all numbers as floats. The format of TI floats is 1 byte for the type (this contains other information such as if the number is negative, or if it is part of a complex number), 1 byte for the exponent, which is only allowed to be on [-99,99], followed by 7 bytes of binary coded decimal (each digit uses 1 nibble, so 14 base 10 digits).

So each float uses 9 bytes. A complex number is 18. In a Real list, each element is 9 bytes, in a complex list, each element is 18 bytes. In a matrix, it can only have real elements and so each element is 9 bytes.

So to answer your question, each element takes 72 bits, but you can only realistically fit about 40 bits of data in each element.