It's grams, but I like this idea. As soon as I find my pocket scale, I will give this a try in slic3r. 

Have you measured the wall thickness after calibrating in this way? I have numerous concerns, but I want you to tell me more about your success before I express them.

Sorry for the slow response. The numbers I used are in grains, not grams, so you'd have to convert to grams if you prefer that unit of measure. One thing I forgot to mention is you have to set the flow to 100% to print the test cube, otherwise it screws up the calculations. I haven't done before and after measurements of wall thicknesses, but just from eyeballing it, if the flow rate is too high, the plastic squishes out, so I'm pretty sure it'd end up thicker if some of that was going on. As near as I can tell, manufacturing tolerances on filament are loose enough that absolute perfection is unobtainable, but so far I've found this approach gets me in the ballpark of where I need to be without doing a lot of hunting and trial and error.

Have you measured the wall thickness after calibrating in this way? I have numerous concerns, but I want you to tell me more about your success before I express them.

1. At the bottom of the "basic" menu settings in Cura, there's a box for filament diameter and flow rate. For diameter, enter whatever diameter filament you're using, 1.75 or 3.

2. Build a STL for a box with a volume of 1 cubic centimeter. For example, 1 cm x 1 cm x 1 cm, or 1 cm x 2 cm x .5 cm, etc. That will be your calibration file.

3. You can get digital scales for measuring small amounts of weight starting around $25. They're commonly sold for measuring powder for reloading ammunition, scientific stuff, etc. I calculated the weight of 1 cubic centimeter of ABS at 16.1 grains. That number can be converted to whatever unit of measure you prefer, ounces, miligrams, etc.

4. Print the 1 cm calibration file and weigh it.

5. Divide 16.1 grains by the result, which will give you the percent value to enter in the "Flow" box in Cura. For example, if the weight of the calibration model is 16.5 grains, divide 16.1 by 16.5 to get .9757, or 97.6%.

Manufacturing tolerances on fliament diameter can vary a bit. I've been using this method when I load a new spool of filament and it seems to give good results. It only takes a few minutes to print a 1 cm box and the method is reasonably idiot proof, without a whole lot of fooling around. When I print a second box and weigh it after doing the above, the margin of error in the 16.1 target weight is tiny, usually within 1/10th of a grain.