strange behaviour for VCR

Hello, (sorry for my english)

I was writing a frontend for ngspice-xspice above netbeans in java
for educational purpose (I'm teacher) and I was interested by a
more interactive simulator. So I decided to test gnucap. I am satisfied
with it but also concerned by strange behaviour above xp and Linux.
I'm using gnucap-0.35

I don't know if it is a bug or if I misunderstand something. So please,
show clemency, I spend a lot of time to delimit this strange behaviour.

Here is the smallest offending file : a differantial amplifier with a 
Voltage
Controlled Resistor : http://jacques.lecoupanec.free.fr/img/gnucap.jpg

-------------------------------------------->
GnuCap File

V4 4V 0 4.0
V5 5V 0 5.0
V15 15V 0 15

RC1    15V    col1    100
Q1    col1     4V    em    NPN 
RC2    15V    col2    100
Q2    col2     5V    em    NPN 
I1    em    0    DC     0.001

E10 10 0 col1 col2 100000
I10 11 0 0.033
G11 11 0 VCR 0 10 1.0

*V30 30 0 1000.0
*V30 30 0 10.0

.model NPN  NPN (level=2 BF=500)

.print tran + V(col1)
.print tran + V(col2)
.print tran + V(10)
.print tran + V(11)
<--------------------------------------------------------

If I try without the line "V30 30 0 1000.0", I obtain strange result :
the voltage of the node 11 is 32.77E+12.
If I uncomment this line, I have good result for the VCR. (329.34 V
which is the theoretical result). This Voltage source would be useless !

Perhaps it's a stupid error but I don't understand...

Best regards

Jacques Le Coupanec