Why is the mutual inductance term negative in the loop equations?

[xxXen0xx]

http://img638.imageshack.us/img638/989/newbitmapimagety.png
I am running my loop current equations for this circuit and have.
-1+-j6*(I1-I2)=0
-j6*(I2-I1)+j20*I2 +j12*I3=0
(j30-j2+j5+4)*I3+j12*I2=0

But i am being told that the mutual inductance term j12 should be negative in both.
My model current I2 is going into the dot and I3 is coming out of the dot, so
jwL+M=0 should apply
not
jwL-M=0

what am i missing?
o wait... is it the sign of the 1V i have backwards?

 

[technician]

what is the question?

 

[xxXen0xx]

I am reviewing questions from old studies.
I believe it was to find Zeq at the terminals of the left hand inductor, by using a 1V test.
But i think I answered my own question.
For I1 running from - to + potential the loop equation needs to read
1+-j6*(I1-I2)=0
so that second loop equation is
-j6*(I2-I1)+j20*I2 +j12*I3=0