We present the results of analyses of the ground level enhancements (GLEs) of cosmic ray (CR) events of 29 September 1989; 15 April 2001 and 20 January 2005. This involve examination of hourly raw CR counts of an array of neutron monitors (NMs) spread across different geographical latitudes and longitudes. Using awk script and computer codes implemented in R-software, the pressure corrected raw data plots of the NMs were grouped into low-, mid-, and, high-latitudes. The results show both similarities and differences in the structural patterns of the GLE signals. In an attempt to explain why the CR count during the decay phase of GLEs is always higher than the count before peak, we interpreted all counts prior to the peak as coming from direct solar neutrons and those in the decay phase including the peak as coming from secondary CR neutrons generated by the interactions of primary CRs with the atoms and molecules in the atmosphere. We identified NMs that detected these primary neutrons and found that they are close in longitudes. Previous authors seemingly identified these two species as impulsive and gradual events. Although there are a number of unexplained manifestations of GLE signals, some of the results suggest that geomagnetic rigidity effectively determines the intensity of CRs at low- and mid-latitudes. Its impact is apparently insignificant in high-latitude regions. Nevertheless, the results presented should be validated before making any firm statements. Principally, the contributions of the ever-present and intractable CR diurnal anisotropies to GLE signals should be accounted for in future work.
