Little endian: byte N has value 256^N, bit n has value 2^n.

Big endian: byte N has value 256^(L-N), bit n has value 2^n or 2^(l-n) depending on the architecture (some effectively have little bit-endian but big byte-endian) and where L and l are the byte and bit size of the whole integer respectively.

Design hardware or even write arbitrary precision routines, and you'll quickly realise that "big endian is backwards, little endian is logical".