Accounts do not survive but Greek astronomers must have used an approximation method based on interpolation, which is still used by some amateurs. This method consists of recording the declination angle at noon during some days before and after the solstice, trying to find two separate days with the same declination. When those two days are found, the halfway time between both noons is estimated solstice time. An interval of 45 days has been postulated as the best one to achieve up to a quarter-day precision, in the solstice determination. [ 39 ] In 2012, the journal DIO found that accuracy of one or two hours with balanced errors can be attained by observing the Sun's equal altitudes about S = twenty degrees (or d = about 20 days) before and after the summer solstice because the average of the two times will be early by q arc minutes where q is (πe cosA)/3 times the square of S in degrees (e = earth orbit eccentricity, A = earth's perihelion or Sun's apogee), and the noise in the result will be about 41 hours divided by d if the eye's sharpness is taken as one arc minute.