S U M M A R YExperiments on harvesting sweet potato as a green vegetable and as a root crop are described. Whole shoots yielded 62% more than shoot tips. Similar total shoot yields were harvested whether tip removal was at two, three or four week intervals. Root yield was decreased by 31 to 48% by removing shoot tips, while removing whole shoots led to root yield decreases of 48 to 62%. Harvesting shoots at two week intervals gave 72% reduction in root yield, compared with 50% with four week intervals. There were fewer and smaller tubers as the frequency of shoot harvests increased. There were varietal differences in response to shoot removal For reasonable yields of both shoot tips and tuberous roots harvesting shoot tips at four week intervals is recommended.Although the sweet potato is generally grown for its starchy tuberous roots, there are many people in Africa and Asia who eat the stems and leaves (Villareal et al., 1979a). The foliage, which is of good nutritional composition (Kay, 1973), is also used locally to feed stock.The use of shoots as a vegetable or as forage necessitates some shoot removal while the plants are growing and might be expected to decrease the supply of photosynthate to the growing tubers. Villareal et al. (1979b) and Gonzales et al. (1977) reported that tuberous root yields were generally decreased by frequent harvesting of the tips.These experiments were therefore designed to investigate the effects of the method and frequency of de-topping on the yields of shoots and roots. METHODSTwo separate experiments were carried out, one in 1977 and one in 1978, at the International Institute of Tropical Agriculture (IITA), Ibadan, Nigeria. In 1977 two sweet potato varieties, TIS 2328 and TIS 2154, were subjected to three patterns of shoot removal. A factorial (3 X 2) design was used, with the treatments arranged in four randomized complete blocks. The patterns of shoot removal were: (i) no removal; (ii) shoot tips removed when about 25 cm long and (iii) most of the shoots removed to within 10 cm from ground level. Each

SUMMARYThe relation between source potentials and sink capacities of four cassava (Manihot esculenta Crantz) varieties with different characteristics was investigated by means of reciprocal grafts. There were marked differences in the sink capacities (average stock effects) and source potentials (average scion effects) of the varieties.

S U M M A R Y Total fresh leaf yield of cassava (Manihot esculenta Crantz) variety Isunikankiyan was not significantly affected by harvesting frequency, unlike variety TMS 30211. Compared with plants whose leaves were not harvested, there was a decrease in total fresh tuberous root yields of 56-76%, 34-62% and 15-32% when leaves were harvested at 1-, 2-and 3-month intervals respectively. The two cassava varieties reacted differently to leaf harvests in terms of tuber yields. Harvesting the top leaves of cassava at 2-or 3-month intervals is recommended to ensure reasonable yields of both leaves and tuberous roots.The tuberous roots (tubers) of cassava (Manihot esculenta Crantz) provide the basic food for many people in the tropics but the leaves are also a popular food for many African peoples, whilst there is an increasing interest in their use as a livestock feed (Cresswell, 1978). Cassava leaves are a good source of protein, vitamins and minerals. Oyenuga (1968) reported that local Nigerian varieties averaged 14.7% protein, 8.4% ether extract and 16.1% total ash and Eggum (1970) found that the amino acid content of leaves of three Nigerian varieties averaged 6% lysine, 2% methionine, 11% aspartic acid, 6% valine, 5.5% arginine and 2.2% tryptophan. Cassava leaf meal is nearly equivalent to alfalfa meal in feed value (Khajarern et al., 1977).However, Ahmad (1973) and Singh and Chaudhury (1975) have reported adverse effects of leaf harvesting on yields of tubers and it seems that methods must be developed to allow for harvesting cassava leaves while at the same time maintaining reasonable tuber yields. An experiment was therefore conducted to investigate the effects of frequency of leaf harvests on yields of fresh leaf and tubers, yield components and dry matter in two cassava varieties and the results are reported herein. MATERIALS AND METHODSLeaves from the top 30 cm of each branch of two cassava varieties, Isunikankiyan and TMS 30211, were harvested by plucking them from the stems at various intervals of time. A 2 x 4 factorial design was used with treatments arranged in four randomized complete blocks, one factor being the two varieties and the other the timing of leaf harvests, namely no harvest or harvests at % Present address: Njala University College, PMB, Freetown, Sierra Leone.

Gari and foofoo (fermented cassava foods) were purchased in the Freetown markets of Sierra Leone and analysed for factors associated with quality. Sellers were generally knowledgable about the source of the products which all originated from the country areas outside Freetown. The mean cyanide contents were higher (8.6 mg kg-' DM for gari and 28.2 mg kg-' DM for foofoo) than the amounts suggested by the Codex specification and, because of the distribution of the values around the mean, some samples necessarily contained unacceptable amounts of cyanide. Cyanogens were present as the cyanohydrin or as free cyanide; no glucoside was detected. Microbiological analysis of the samples showed high counts for total organisms, fungi and Enterobacteriaceae ( lo6-lo7 g-'). The mean water content of gari was 119 g kg-' which falls within the recommended limit of 120 g kg-' but again, due to the distribution of values around the mean, some samples had considerably higher water contents than that limit. Titratable acidity and pH were measured and the mean pH values were 4.18 for foofoo and 4.55 for gari. The particle size and swelling power of gari samples were also measured as these relate to consumer acceptance. The data show the quality of current cassava foods marketed in a major urban centre in Sierra Leone. As centralised processing develops, improvements can be. made to those quality parameters which have been identified as important.

The preparation and distribution of foofoo comprises several stages, namely soaking, dewatering and storage prior to sale. The effects of four different time/temperature soaking processes on the quality of foofoo (pH, microbial load, residual cyanogens) were determined. The processes were chosen as they gave roughly equivalent degrees of retting (the softening of cassava during soaking) by which the progress of the initial fermentation is normally judged. Instrumental measurement of cassava hardness confirmed that the processes were equivalent within experimental limits. The four soaking processes had no significant effect on the water content of foofoo, nor did this change during storage. Processing did affect the pH, total titratable acidity and microbial profiles. Analysis of the aqueous extracts from the processes showed that the glucose present in the fresh cassava was fermented into a mixture of acetic, butyric, isobutyric and lactic acids with some ethanol. The lowering of the pH reduced the number of Enterobacteriaceae to the limit of detection and also affected the types of microbe present. The different processes also affected the removal of cyanogens from the product and the processes at 20°C and 35°C led to unacceptable levels of cyanogens (mainly cyanohydrin) in foofoo.