An interesting observation made by an American biologist H.C. Bumpus (1899) provides a good explanation for normalising selection. Bumpus collected some 136 injured house sparrows after a severe snow and sleet storm with high winds at Woods hole. Of these birds 64 died, resulting in two groups of sparrows, those that were killed by the storm and the ones that survived. Bumpus, made measurements of a number of randomly chosen traits such as wing length, wing span, tarsus length etc and found that those that were killed by the storm had measurements which fell at the ends of the bell shaped curve. In other words, birds which had either mean or close to the rnean measurements were the ones to survive. Normalising selection generally eliminates during a catastrophe or a stressful situation individuals whose traits vary markedly from the mean values. By way of interpreting Bumpus's observations, it could be said that the birds which were blown down easily by the winds had either their wings too long for their body weight and therefore presented a larger surface area or they had too short wings for their body size and therefore could not fly against strong winds. Prjor to the catastrophe the range of individual measurements was larger as compared to the range of measurements that survived the catastroihe. Every catastrophe may narrow down the variability of the population thereby weeding out suboptimal genotypes.