182 MARION EXPEDITION TO DAVIS STRAIT AND BAFFIN BAY 



The correlations publishetl as a result of these investigations at 

 the British Meteorological Office in 1925 (Smith, 1927, pp. 45^8), 

 lacked an actual count of the icebergs from 1900 to 1906, but in its 

 place the value of the atmospheric pressure gradient Ivigtut to 

 Belle Isle was substituted. As soon as access was had to the files of 

 the United States Hydrographic Office a new table of iceberg values 

 was constructed for this gap of seven years, and correlations were 

 recalculated. The results of this work changed slightly the values 

 of the correlations based upon the original variates. 



The next task of the investigation centered upon a correct arrange- 

 ment of the iceberg data, the period 1880-1926 being divided by years 

 into six groups depending upon the relative number of icebergs 

 ])resent in the western North Atlantic south of Xewfoimdland. For 

 example, the lightest iceberg vears were found to be 1881, 1889, 1893, 

 1900, 1902. 1917. and 1924, and the heaviest iceberg years, 1903, 1909, 

 1912, 1914, and 1915.'^ The six groups were next arranged in ac- 

 cordance with a scale to 10, but it was found by plotting a dis- 

 persion diagram that the iceberg values failed to follow the require- 

 ments of a variable ; that is, the greatest number of deviations did not 

 fall proportionately near the mean average. The fact that a value 

 either too great or too small had been assigned to the iceberg years 

 showed that too great weight had been given the extremes. In order 

 to correct this error, it was necessary to arrange the years in con- 

 formity witli the shape of a probability curve, and then from a trans- 

 formation curve to obtain fresh relative values of the icebergs. The 

 following table gives the values weighted according to scale to 10 : 



The meteorological material was next arranged in a form con- 

 venient for correlation with the iceberg values, which was made 

 possible by the material and the assistance furnished by the British 

 Meteorological Office. The best comparison between a series of 

 years or months is afforded, not by pressure values direct but by the 

 departures from normal. Station normals were obtained by arrang- 

 ing a long series of observations, and with the aid of these data 

 isanomaly maps were constructed for the months October to March, 

 for the years 1880-1926, inclusive. A classification Avas then at- 

 tempted based upon a grouping of the monthly anomaly maps, 

 October to March, for the northwestern North Atlantic region, from 

 which it was apparent that two types of pressure distribution stood 

 out clearly, one showing an excess of pressure centered in the region 

 of Iceland and more or less dominating the ocean ; the other, show- 

 ing a dominance by the reverse condition. The first or "plus" 

 type is subject to further classificaticm into gi-oups 1 and 2, depend- 

 ing upon a i-elative intensity of the excess mass of air. Combination 



'"The year 1!»2'.), with over l,;iOO bergs south of Newfoundland, exceeds anything on 

 record. 



