The Variation of Latitude

Data of the International Latitude Service from 1899 December to 1967 July are analysed by two methods. The annual terms are estimated from monthly means over 7-yr intervals; there are also variations of means over these intervals, apparently irregular, of order 0.05^″, but in the coordinate m( = – y of the I.L.S. reports) a drift started about 1945 and has now reached about – 0.2^″. This is contrary to any hypothesis involving a uniform rate of drift. The means were interpolated; the results and the annual terms were subtracted from the data, and the remainders were analysed harmonically over 14 month intervals. The first method was by comparison of expectations over ranges of 1, 2, 4, 8 and 14 times 14 months, but was indecisive except in showing that contribution from observational error was probably small. The second used maximum likelihood, with a small correction for observational error. The free period is found to be $$434.3\,\pm\,2.2$$ sidereal days; the time of relaxation has a large uncertainty, the most probable value being about 23 years, but values from 14 to 70 years would lie within limits corresponding to the standard error.