For a long time I used to read the hydrometer at the bottom level, but then I read Designing Great Beers by Ray Daniels (p. 14), where he says that he has changed to read at the top of the meniscus after having misread it for nearly six years (myself at least that long). At first I was skeptical, but by taking a reading with tap water at the calibration temperature it was clear that the 1.000 line is at the top of the meniscus, not the bottom. Reading at the bottom of the meniscus will over-report the specific gravity by 2-3 points.
I had been confused by how I was taught in chemistry class; "glass instruments such as burettes should be read at bottom of the meniscus". Here's the reason I think beer hydrometers have been designed to read at the top of the meniscus. The hydrometer scale is viewed inside a test tube with wort, whereas a burette has its scale printed on the outside with liquid within. The wort is often opaque so that we cannot see the scale through it. By reading at the top we can get an accurate reading of the scale, because it can be seen above the opaque liquid.
I invite everyone to test water (tap, well, distilled, the s.g. difference is negligible) at the calibration temperature of your hydrometer and see where the 1.000 line is in relation to the top/bottom of the meniscus.
By the way, the image showing the liquid and hydrometer is missing the upward curve of the liquid at the test tube. It should curve up there just like it does on the hydrometer body. --Quaffer 13:12, 3 February 2010 (CST)