Since you bring up the subject of other assemblers, the previous
    answers stating that the '$' stands for current PC aren't exactly
    correct. It refers to the assemblers' "Location Pointer".

Bless you for pointing this out.  I was about to do so in a rather more
frustrated language!

    In a PIC they are the same thing when used in program memory as all
    instructions are 1 word. In other micros where an instruction can be
    more than 1 byte, it is the address (PC value) for the first byte of
    the instruction.

The assembler does not have a PC.  In a PIC, the actual value of the PC
during execution is more or less irrelevant, because the PC is not directly
accessable to the programmer (right?)  However, on a processor where the PC
is a a program accessable register (say a pdp-11 :-), there is FREQUENTLY a
difference in PC and <assembler-assigned-value of "$" or equivilent>,
because the PC is usually incremented during instruction and argument
fetch, before the actual execution.  Thus, on a pseudo-PDP11:

        A:      jmp $+1
        B:      add PC, 1
        C:      add ...
        D:      sub ...

The jump instruction at A will jump to B, but the jump at B will jump to D
because PC has already incremented to C before the add occurs.  This is the
same thing that results in the "jmp $-1" confusion.  If you're doing actual
math on the program counter, it's a single instruction loop.  If you're
using an assembler that defines $ as the current instruction, it's a two
instruction loop.

BillW