Cliver & Ling 2016, "error testing" group number: description.
Test reconstruction using the original Hoyt & Schatten (1998) with corrections.
This series was obtained by reproducing the method described in the original publication (Hoyt & Schatten 1998) with corrections of specific defects. After matching first exactly the original values as a validation, the calculation was repeated after removing the initial trend in the scale of the Greenwich group numbers used as the starting calibration. Except for this correction to the data, the rest of the k coefficient determination remained unchanged. After checking two main possible mechanisms by which this scale inhomogeneity could affect the numbers before 1874, the drift in the Greenwich data proved to be the prime cause of the 40% underestimate of the early part of the original group number series.
As this series was built mainly to test the 1880-1915 "Greenwich" trend, it was only computed back to 1841 and up to 1976. The current files do not include standard errors.