Very strange behaviour with LCD display

[See_Mos]

18F1320  It's an old device but I have a few and they are ideal for what I need at the moment.

I have two PCB.  Both have sockets for the 18F1320, a 6 pin connector for the PicKit and an LCD display.  The only difference is the display.  The fist is 4 lines by 20 and the second 2 lines by 20.  Both are wired the same using RB4....RB7 for the data lines.  The fault arose after the failed 2 line display was replaced with a new one.

Here's the puzzle.  I can program a device in the 4 line board and transfer it to the 2 line board and it works as expected.
 
I cannot program a device in the 2 line board unless I disconnect RB4 and RB5 from the display.  Once programmed the device will work in both boards.

The 2 line display is a new RS 532 7470 that I bought a while ago.

[trastikata]

It is not the PIC, it is the LCD which has low impedance on those pins, can you measure it?

[Giuseppe]

It could be the power supply of the programmer not holding with the display connected.
this if you power the board via the programmer

[tumbleweed]

Does the display produce any voltage on the RB4/RB5 lines when you power it up?
If it does, that could be upsetting the programmer (voltage backfeeding to VDD/MCLR through protection diodes).

[See_Mos]

Obviously it is the display causing the problem as the board was OK with the old display and the spare is OK with the 4 line display.

The measured resistance from the display pin to pin and pins to supply rails is around 700K.

Tumbleweed has the answer, I removed the PIC from the socket and powered the board from the PicKit.  VDD is 4.65V, RB4 and RB5 are 4.6V.  RB6 and RB7 were only 0.2v because Pickit 2 has internal 3K resistor to VSS.  I added 10K pull down resistors on RB4....RB7 and and the voltage dropped to 0.77.  Programming now works so problem solved.