with two maize varieties and two levels of fertilizer (0 and 30 kg N/ha)" (Hildebr and and Poey, 1985:127-8). Since area farmers intercrop, all the treatments had a mix of maize, cowpeas, and sunflowers (Hansen, 1986).
It was specifically requested to the village headmen and the extension workers who selected the farmers that half of the cooperating farmers be women. However, only 40% of the cooperators selected in one village and 30% in the second village were women. In a number of cases the women and the men farmers selected were not comparable as farmers. The men tended to be vigorous individuals in their middle years and many were high resource farmers who owned cattle. Most of the women tended to be low resource, older individuals at the end of their life cycle. They were probably selected because age is revered and it was considered an honor to be selected, perhaps more so for women than for men. Therefore, comparisons between male and female-farmers in terms of management and yields would not be valid to show gender differences in farming (Table 8).
The data do show differences between high and low resource farmers and more women are in the latter category. A modified stability analysis was carried out on these data by Hildebrand (Hildebrand and Poey, 1985:126-34) and by Hansen (1986). Because of the inclusion of a range of farmers, young and old, male and female, an evaluation of the types of evironments could be made where "environment.o.becomes a continuous quantifiable variable whose range is the range of yields from the trial" (Hildebrand and Poey, 1985:126). The analysis showed that in the same area there were a range of environments in terms of farmer management, soils, rainfall, and the like, and that the cultivars respond differently. The local cultivar was superior in poor maize environments while the improved maize was superior in "good" maize environments (Tables 9a and 9b). Both cultivars responded "favorably to fertilizer in both good and poor environments" (Hildebrand and Poey, 1985:129). The data showed that there were two different recommendation domains. Although there were both men and women in each domain, there was a tendency for the women farmers to be in the poorer environment most likely because they were low resource farmers to begin with. Further analysis using confidence levels allowed the high and low environment farms to be compared. The results showed that only farmers in the better environments should choose the improved variety (the composite) and that they should fertilize the crop (Table 10a). In the poorer environments, the local variety was better (Table l0b). Fertilizer helped, but should only be recommended to farmers who could afford it. The final recommendation was "to fertilize the local maize variety in the poorer environment and to use the composite maize with fertilizer in the better environment" (Hildebrand and Poey, 1985:132). It should also be mentioned that farmers who owned cattle and used the manure on their fields were in the better environments. Women did not own cattle too frequently, although the one high resource farmer in the sample did.
By comparing people at different ends of the spectrum, two recommendation domains-were discernable. All but one of the farmers in the better environment were men. Most of the farmers in the poorer environments were women, although there were some men. This showed that recognizing different segments of the population, including those at particular risk, resulted in the delineation of multiple domains and technology solutions in this case. The female headed households constrained by labor and cash would find it difficult to use fertilizer and this coupled with their smaller holdings and lack of extension advice would make their use of the improved cultivar