Carbon-14 dating should work in principle, but in practice the error on carbon-14 dating is several decades. So that's out.

I'm not sure about tritium dating for the water. Reading the USGS website on it, it seems to indicate that the way tritium dating works is that they seal some water in a copper tube and see how much helium magically appears as the tritium decays. They ask for two 500mL (two 1 pint) samples.

So if tritium has been decaying into a sealed glass container for 40 years, should there be a measurable amount of helium in the container?