calculated vs measured LJP

I am using two different intracellular pipette solutions and one bath solution (standard ACSF) to measure NMDAR-IV curves in CA1 pyramidal cells. I am trying to correct for the liquid junction potential of my two different internals, but the values I calculated using Igor (10mV and 8.4mV) are very different from what I've measured on my rig (26mV and 22mV). 

Please advise on which numbers I should use to correct my membrane voltage, and what could be causing such a large discrepancy between experiment and theory?

(I am using a Ag/AgCl pellet as ground, which should be fine since I'm not changing extracellular Cl.)

Thanks!!

 I am just curious to know how you calculated the LJP.
Would you provide the composition of both intra and bath solutions? I understand Igor is the software, not program.
How you measured the LJP? By changing the bath solution to the another one? Do you have the agar bridge between recording chamber and the reference electrode? 
Thanks.

Hi guys, I would like to ask a similar question.
I used JPCalc to calculate the liquid potential for my experiments, in order to verify the result I measured the liquid junction potential as well.
The problem that occurs is that when I change my aCSF bath solution for my internal solution the potential I measured before does not reverse, which it should.

Can somebody  give me some advice on this?

My measurement procedure is as follows:
I start with internal solution in my chamber, my bath electrode is a Ag/AgCl pellet that is located in an external bath, containing internal solution. This external bath is connected to the recording chamber via 3% Agarbridge (3M KCl). When entering the internal solution in the recording chamber with the pipette I off-set the measured potential, then I change the internal solution in the recording chamber for aCSF, I also exchange the agarbridge. The measured liquid junction potential is very close to the one calculated with JPCalc. To check for the reversal of the measured potential, I exchange the aCSF in the recording chamber with internal solution and also switch the agarbridge. Theoretically, the measured potential should reverse, but instead the value increases.
I have no idea what is causing this effect.

 Ok, I have stopped using the agar bridge and instead started using a flowing KCl bridge. With this approach it worked fine. 

I still don't understand why my agar bridges did not seem to work correctly.