When considering variations of the surface temperature of the sea it is common practise to concentrate on the mean values over a certain period. It is argued that for some purposes the extreme values may be as relevant. The hydrographic conditions in Faroese waters are dominated by the presence of the Polar front (fig 1—2). A unique series of daily surface temperature observations exist from Mykines. The long-term variations have been given by Smed (Smed Ann. Biol., fig 3) based on mean monthly values.
 In this paper we have looked for deviations from the mean on a shorter time scale i. e. by days. A visual inspection of the data (fig. 4) indicates that the deviations are not symmetrical but rather more negative. An analysis confirms this hypothesis. The data were high pass filtered and the number of deviations of various temperature intervals were counted. A distinction was made between positive and negative deviations. Table 1 and fig. 6 demonstrate the result. The temperature variations atMykines, thus are interpreted as composed of three parts: 1 A yearly variation, 2 »Noise« symmetrical with respect to the running mean, 3 Negative anomalies. It is the latter which has been studied in more detail. Figs. 7 and 8 demonstrate the distribution of anomalies ( > V Í ° C) through the years (1920—1940) and in various months. The anomalies are significantly more frequent during wintertime.
 As for the origin af the anomalies, it is argued that these are probably of an advective character indicating the presence of cold »blobs« or »eddies« in the waters of the Faroe shelf. Further evidence for this is presented in fig. 10 and evidence that temperature variations in these waters include advective processes as a dominant part is found also from a series of temperature ■— current observations from SuSuroyarfjørður (figs. 11, 12, 13, 14). The temperature variations observed were plotted against the Lagrangian coordinate x(t) of a fictitious water particle
 tx(t)=2v(i)-/\tKv(i) being the observed current along the main axis at the time i and A t , the sampling interval. The fact that in these plots the curves are very often retraced when the current turns — as exemplified by fig. 15 ■— indicates the importance of advective processes. As to the origin of the eddies, we assume their creation to be related to the Polar front which indeed does produce such entities (B. Hansen & J. Meincke, to be published). In conclusion we note the importance of such processes both from ahydrographical and biological point of view.