36 DISCOVERY REPORTS 



The steep fall from i September to i November and the sharp rise from i November 

 to i February are both at variance with and opposite to the known effect of phytoplankton 

 production in situ. In the southern region there is no effective production before the 

 first week in November, and then the concentration and the gradient of the increase 

 are small. This phytoplankton production reaches a peak in the third week of February 

 and falls slowly to the third week in April, when a small autumnal rise takes place until 

 the end of the first week of May. Therefore the fall in silicate content from September 

 to November is occurring at a time when no phytoplankton is being produced; 

 similarly the rise of silicate from November to February is taking place at a time 

 when silicate is being utilized by a production of phytoplankton. Only the flattened 

 gradient of the rise in February agrees to some extent with the peak period of phyto- 

 plankton concentration. The fall of silicate in March is occurring at a time when the 

 phytoplankton concentration is slowly diminishing. 



During the circumpolar cruise of the southern winter of 1932 it was noticed on at 

 least two occasions that there was no movement towards the north across the Antarctic 

 convergence of Antarctic surface water. There was, on the contrary, evidence for a 

 southerly flow of warmer water which is probably a winter feature explained by the 

 fact that the northerly flow in summer, having its origin in melting ice to the south, is 

 reversed when new ice is being formed near the coast of Antarctica in winter. Such 

 a movement in winter towards the south may account for some of the fall of silicate as ' 

 shown by the southern curve in September and October. However, such a southerly 

 movement cannot be the sole reason for the fall in these months ; later in this section 

 another reason will be discussed. Unfortunately we have no averages for June and 

 July, so that the extent of the fall in silicate content before 1 September cannot be 

 known. The average values at 1 May and 1 September differ by over 800 mg., a 

 difference which gives some indication of the extent of the fall of silicate content during 

 the intervening months. 



The rise in silicate content from 1 November is steep until 1 February, when it 

 flattens during February and falls in March. The flattening in February and the fall 

 in March may represent the effect of the peak concentration of phytoplankton 

 production in situ for the southern region; if so, the effect of production in situ has 

 only checked the rate of the rise of silicate content which started in November and 

 has not been enough to override the increase, since in April the curve is again 

 rising. The rise of silicate over the months of November, December, January, 

 February and April with the evidence of masking in February and March must be 

 due to the arrival at the southern region of warm deep water with a high silicate 

 content some time previous to 1 November; this warm deep water then transfers its 

 high silicate content to the Antarctic surface water by mixing. The rise over all these 

 months cannot be due alone to regeneration in situ because the rise starts in November 

 when the concentration of phytoplankton in the southern region is small, and only in 

 late February does the production come to a peak. The possibility that the warm 

 deep water carries back to the Antarctic zone undissolved diatom skeletons which 



