The big/little endian property applies to more than one level.
The most common "problem" is the byte order.
An example, the value 0x1234 (hexadecimal in C-style)
The bytes are stored in the following order in a little-endian machine: 0x34 0x12
In a big-endian machine. 0x12 0x34
It is easier for a human to read a big-endian byte hex dump, but from a hardware/software standpoint I prefer little-endian.
The we have endianness on the bit level (bit numbering within a byte/word).
The strict big-endian machines, like PowerPC, say that the highest bit is bit 0.
I think that is crazy. I can't see any advantage at all, only problems.
There are mixed-endian architectures like the 68000 serier. Big-endian in the byte level but "normal" little-endian bit numbering witing words.
The Intel processors used in PC's are little-endian.