How far does IR penetrate Water? (failed?)

An attempted experiment to measure how far IR penetrates (tap) water (UK variety)

Problem points:
  • The wide necked Vacuum flask turns out to be a wide necked flask!
  • The high frequency cut off of the IR pass filter is not known (this was purchased from Edmund Optical)
  • The High intensity Low votage Halogen lamp is not the sun and has a peak output at the red end of solar spectrum (3500K cf 5000K)
Test setup

Thermocouples are placed away from the light input at distances below the water surface of
  • 1mm
  • 3mm
  • 11.2mm
  • 18.5mm
  • 39.5mm
Tube and filter 50mm diameter

2 runs made with and without the IR pass (visible stop) filter:
These produce 2 outputs:

The Filter
Using a camera as a lightmeter and looking at the halogen source the filter reduces the exposure (mainly visible) by a factor of 110.
The light source.
20 watts = perhaps 10 watts into tube and 1 watt (unfiltered) into water

What they show
The spectrum of the bulb is not generated from a hot enough filament. The heating effect of filtered and unfiltered light is very similar.
Unfiltered heating does not predominate at greater depths (it is virtually the same as IR "only")
At a 39mm depth heating effect has little effect over the leakage throught the faulty vacuum flask.

What I would have thought.
Despite the limitations I would have expected significantly greater heat imput (faster temp rise ) at depth with unfiltered light.