When a dual extruder for 3D printing is available it is very desirable to be able to print using a support material that can be dissolved instead of mechanically having to remove it.  I saw an interesting video describing a process that started this investigation.  Here is the test result:

Video:
3DSystems has a very useful video on YouTube about how to use caustic soda (Sodium Hydroxide NaOH) to dissolve PLA used for support.  217 grams is added to 3 liter water used at 60C for 3 hours in an ultrasonic cleaner.  217g in 3 liter is equal to 72.3g in 1 liter is equal 1.8 Molar strength.
http://www.youtube.com/watch?feature=player_embedded&v=uWYn5wJ72zY

The YouTube listed cost of the ultrasonic cleaner is ridiculously expensive.  A few days ago I bought a similar but larger high-power 6l industrial ultrasonic cleaner for $279.
http://www.amazon.com/Commercial-Kendal-Liters-ULTRASONIC-CLEANER/dp/B002QD624C/ref=sr_1_2?ie=UTF8&qid=1346692474&sr=8-2&keywords=kendal+ultrasonic+cleaner

Samples before test
This time I tested several different plastic types and varying strength Sodium Hydroxide.
1.0 Molar: 40g per liter water
1.8 Molar: 72g per liter water (same as the video)
3 Molar: 120g per liter water
6 Molar: 240g per liter water

Plastic: Red and White 2.9mm filament samples of PLA and small samples of ABS, Polycarbonate (Lexan) and Nylon-66.

All the samples and the dissolving liquids were left in the test tubes and tested at the same time in the ultrasonic tank.

Test Result:
After 5 hours in the ultrasonic tank @ 60°C there was no visual effect or surface hardness change evaluating the samples under a microscope of the different plastic types except the PLA.

Amount of dissolved PLA after drying the residue (see top picture):
1 Molar: 13%
1.8 Molar: 42%
3 Molar: 88%
6 Molar: 56%

Five hours is much longer time than the video but dissolving thin scaffolding will be a lot faster than the solid 2.9mm filament so the estimate of three hours sounds reasonable.  The big surprise was that the stronger mixture was worse so it looks like 3-4 Molar is a good value.  A higher temperature would also further speed it up.

In summary it works and looks like a promising way to build scaffolding for 3D printing.

Here is a link to quickly dissolve PLA if no other plastic is involved:
http://www.vinland.com/blog/?p=68