If this description inspires you to build your own, feel free to copy whatever you find useful. 

Click on a picture for a larger version, these are between 100 and 200 kB.

Description of parts:

• For the housing, I used a small acrylic box. (Actually it was the box in which my ID card for my work computer came.) Its outside dimensions are 90 x 62 x 8 mm (when closed, duh); all edges are bevelled. But one could also use a CD (or CD single) box, audio cassette box etc. 

• I glued a thumb-tack (push-pin) on the inside, its pin shortened so as to JUST fit. I also amputated a part of the head so that I could fit in a slightly longer pointer. 

• The pointer is cut out of a spare vertical shades element I had lying around. But any stable material will do. The black line is just for visual clarity. I drilled a 1.0 mm hole. 

• From the same material I made a spacer "ring" so that the pointer would run free.

• The dial (in % and degrees) is a print-out of a high-resolution tiff file (zipped as inclino2.zip, 58 kB). You may print it to the appropriate size. I had to cut a few indents to match the inside shape of the box. 

• The conversion table is put in the box back-to-back with the dial. It is a left-over from the first prototype, which had a scale in degrees only.

The finished product, lying flat in my hand. Until now, the device has suffered no ill effects from being shaken (not stirred). The pointer will rattle in the box but so what?

The inclinometer measuring the slope of the keyboard of my old trusty Hewlett-Packard 41CX calculator. It appears to be exactly 10%, or about 6 degrees. 

Note the adhesive tape at the left-hand side with which the box is sealed. 

Sighting of the inclinometer in the wild, in its natural habitat. The slope of the gravel trail at this point is 27% and it is one of the steepest trails that I have succeeded in riding up a few times. 

The problem with the device being so small is that the result is quite sensitive to exact positioning on anything but the smoothest surface. I usually make a couple of measurements, eyeballing that the device is parallel to the slope "at large", and then roughly average the results. Alternatively, one could bring a ruler or the like to have a longer basis.

But then, how should degree of slope be defined in the first place? The number usually depends on the length scale one looks at. It's a matter of taste but personally I feel that "slope" should be averaged over 5 to 10 feet. Any variation at smaller length scales would be categorised as bumps etc., while variation over longer lengths is to be considered a varying slope.