RODEWALD recently published brief notes about the temperature measurements of the sea surface made by nine North Atlantic weather ships during the past two decades1,2. His remarks were prompted by the regrettable fact that several of these ocean stations are being discontinued, a decision which will make it much more difficult to monitor the temperature trends in the North Atlantic. He presented1 average annual temperature values for overlapping 5-yr periods, showing a distinctive decrease of 0.56 °C from the 1951–55 period to the 1968–72 period (Fig. 1). Rodewald also provides2 three important maps: the changes in surface temperatures from 1951–55 to 1968–72 for the annual mean, and for the months of February and August. Figure 2 shows his analysis of the February changes. There is a remarkable consistency in the values; for example, at Station D (44°N, 41°W) the surface temperature declined by 1.61 °C in the annual average, by 2.12 °C in February and by 2.10 °C in August. These are surprisingly large changes over such a short time span. Since they are based on a well defined homogeneous record of careful observations there can be no doubt that they are real. The core of decreasing temperature lies along the north edge of the Gulf Stream extension, suggesting a south eastward shift of that current and replacement with Labrador Current waters, especially in the vicinity of the Grand Banks. The largest changes are encountered across the Atlantic between 40° and 50°N.