What's the rationale for considering "0x1.max" as invalid ?

[Basile B. via Digitalmars-d-learn]

0x1.max // exponent expected in hex float
0x1 .max // OK
1.max // OK

What's the ambiguity when it's an hex literal ?