[onre]

Dunno how this is supposed to be worded, but it's because the "pointerhood" is tied to the variable being declared, not the type itself. This becomes obvious when you declare multiple variables at once.

  char* cat, dog;  /* one char pointer, one char */
  char *cat, *dog; /* two char pointers */

[kevin_thibedeau]

The minor problem is that a typedef pointer breaks this pattern:

  typedef foo * FooP;
  FooP a, b; // Both are pointers

Pointer typedefs are misguided but there is no denying that C is inconsistent on whether the '*' is part of the type or not.

[teo_zero]

Right. It would be nice if C allowed types and variables to be separated, we could even defines arrays like this:

  char[8] buffer;

Alas, it's not the syntax Dennis Ritchie settled upon.

[cb321]

The trade-off is one vs. two operator sub-syntaxes or in other words "sub-syntax economy". As it is, C has just one operator expression syntax (and one set of operator precedences/associativities). The "forward applicative" expression syntax is "reused" in the "backward" or "inverse" type declarations.