Most calcium tests work this way:
A strong hydroxide solution (NaOH or KOH) is added to the sample to precipitate the Mg as MgOH2, effectively masking the Mg. The remaining Ca (truth be told, Ca + Sr) is titrated using an indicator dye and an EDTA reagent solution.
Ideally, all the Mg gets precipitated, and all the Ca is left in solution for this test. In reality, some Ca also gets precipitated out as CaOH2 during that first step. As the EDTA titration progresses, there is less and less free Ca in solution, which causes some of the precipitated CaOH2 to go back into solution.
Stopping when you first see the blue color is actually stopping too soon, giving a low reading. For a more accurate reading, wait for the color to change back to purple/pink, and continue titrating until you get a stable blue color.
While we're on this subject, you can achieve a more accurate and sharper endpoint with this test, *if* you already have a good idea what your Ca level is, by adding 90% of the EDTA titrant necessary to complete the test *before* adding the hydroxide reagent [PATE J. B. and R. J. ROBINSON (1958) The (ethylenedinitrilo) tetra acetate titration of calcium and magnesium in ocean waters. 1. Determination of calcium. J. mar. Res. 17, 390-402.]. The EDTA will bind with the bulk of the Ca, making it unavailable to be precipitated as calcium hydroxide. If you choose to do this, pay no attention to the color of the dye until after you have finally added the hydroxide. As long as it is still pink after the hydroxide addition, then you haven't overshot the endpoint yet; just complete the titration to the blue endpoint.
