A few comments:
Cylindrical cells, at least A123 cells, have the can tied to the + terminal (cathode), not the negative.
First thing to do is find out if these cans are supposed to be connected to one side or the other of the electrode stack. The manufacturer has to answer this question.
Next is to characterize their isolation (or lack thereof) at the cell level. You have already done some work to this end, but there are more tests you can run that will give you a better picture of what's going on here. It could be that you are seeing "surface charge" being capacitively coupled to the cans. Measure the cell can with reference to ground with a good quality DMM. Then do the same with reference to the positive terminal (you may see negative voltage here). Pay close attention to the voltage you see immediately when connecting the meter. It may very well drop off sharply within a second or less. (I use the logging feature of the Fluke 289 to make curves of the results--this is a very telling method.) If you get voltage that does not fall off quickly or sharply, try putting a 1M ohm resistor in parallel with the meter and see how the voltage drops. Remove the resistor and measure how/if the voltage bounces back. Often you will see an indication of capacitive coupling between the can and electrode stack. Even though some amount of charge can be transferred to the can in this way, it's not dangerous as-is if the load of the meter (or 1M resistor) is enough to drain off the charge quickly. Current can flow, but it's as though it's flowing through a resistor of very high value, so there are no fireworks when you put cells like this on top of each other, You may, however experience complex self-discharge behaviors in a pack assembled without isolation between cans. Worse yet, if vibration or impact causes a loss of isolation between the can and electrode stack in one or more cells, you can be in for something spectacular. For this reason, you should NEVER build a pack of cells with metal exteriors without insulating the series cell groups from each other. Sudden loss of isolation can be dangerous. Most often, you'll just get leakage voltage onto the pack surface that may tickle you now an then if you can feel it at all. You're more likely to notice one of more cells are experiencing high self discharge because they are "leaking through each other" via multiple leakage paths.
This is another one of those concepts that's kind of hard to wrap the brain around at first, but it's really basic at the end of the day. Hopefully that clarifies the possibilities a bit.
