In the preceding paper in this series (Pearce, 1992) it was explained how an experimenter can ask specific questions about the treatment responses and can obtain answers to them by using contrasts of interest. Here, two standard cases are examined, one in which treatments are quantitative in nature, like the amount of fertilizer applied or the dates on which spraying takes place, and the other in which the treatment set is formed from all combinations of two or more other sets. (Such a ‘factorial set’ might be formed from several kinds of herbicide being used with a range of cultivations.) Finally, a non-standard example is examined in which a well considered set of treatments provided information directly relevant to the subject of enquiry. The subject is considered with some emphasis on testing, but in many instances estimation would be better.

An experiment has its origin in the need to find answers to stated questions. From the start great care is given to its correct conduct (e.g. the application of treatments and the method of recording) as well as to statistical design, always with the original questions in mind. The analysis of its data is the climax of a long process and the questions to be answered must dominate all else. It is not enough to feed data into a computer package in the hope that it will provide an automated path to a true interpretation.

Where the treatments have been chosen with care to answer specific questions, the statistical way of designating purpose is to declare ‘contrasts of interest’, each corresponding to a degree of freedom between treatments. They derive solely from the reasoning behind the selection of treatments. If possible the questions posed should be equal in number to the degrees of freedom and should admit of separate study because no one can give a single answer to several diverse questions.

This paper shows how to define a contrast of interest and how to isolate it in the analysis of variance. Attention is given both to its contribution to the treatment sum of squares and to its variance (i.e. the precision with which it is estimated). Independence of estimation is also considered. Algebraic formulae are given for a restricted though important range of designs, which includes those that are completely randomized, in randomized blocks or in Latin squares, all treatments having the same replication. The methods can, however, be generalized to cover all designs. With these formulae it is possible both to test the existence of an interesting effect and to set confidence limits round its estimated value.

This review gives a historical perspective of the development of Farming Systems Research (FSR) in India over the past 60 years, as India changed from a traditional, subsistence oriented agriculture to one based on science and technology. The first period relates to the years 1930–50, when the Government of India created the Indian Council of Agricultural Research (ICAR) and initiated research on dryland agriculture.

The second period relates to the development of coordinated agronomic trials and simple fertilizer experiments on farmers' fields during the years 1950–65. This development was a forerunner of networks of coordinated applied agronomic research relevant to farmers' cropping systems and adaptive research with farmers' participation. The development of a soil conservation research network in the same period strengthened research on the natural resource base countrywide. The third period started with the reorganization of ICAR, the establishment of State Agricultural Universities and the introduction of coordinated research programmes on high-yielding varieties (HYVs). At the same time the coordinated dryland agricultural research programmes, and subsequently the Central Research Institute for Dryland Agriculture (CRIDA), were established, leading to the development of a national system of FSR for rainfed agriculture.

The fourth period of FSR started with the establishment in 1972 of ICRISAT, an international agricultural research centre that accepted FSR as its mandate and developed the concept of integrating the management technologies for various components of climate, soil, water and crops with a farmers' perspective. ICRISAT's work was complementary to that of CRIDA and helped to bring about conceptual changes in Indian FSR through research aimed at understanding principles and processes of semi-arid tropical (SAT) farming systems. Following this, mechanisms for strengthening on-farm research were emphasized so as to provide stronger linkages between researchers, extension workers and farmers.

Today, FSR with a farmers' perspective occupies pride of place in India's agricultural research agenda. Yet it is a long way from bridging the gap between the generation of technology and its adoption by farmers, who have shown a preference for its components rather than for the full package of technology.

Field experiments during the winter seasons of 1986–7 and 1987–8 studied the effect of three sub-surface drain spacings and three levels of phosphorus on the yield, chemical composition and uptake of nutrients by Indian mustard (Brassica juncea). The number of siliquae m-2 and seed yield decreased with increasing drain spacing. Application of phosphorus increased seed yield and yield attributes. The concentrations of nitrogen, phosphorus and potassium in the seed and stalks decreased and those of sodium, calcium and magnesium increased with increasing drain spacing, but application of phosphorus increased the concentration of these nutrients in the seed and stalks. Absence of phosphorus in the drain water effluent and the level of available phosphorus in the soil profile after crop harvest indicated very slow movement of phosphorus, most of which was retained in the top 30 cm of soil.

The radical concept of potato production from true potato seed (TPS) was adopted as a component of their farming system within three seasons by 23 farmers from Cibodas, West Java. The farmers showed an ability to conceptualize and experiment and desired concepts from which they could develop, through research, appropriate principles and field techniques. Information received solely as detailed practices or techniques hindered their progress, since they first needed to repeat the technique in order to understand the concepts and principles involved. Farmer experimentation resembled closely that of experimental station researchers, with the use of replication in space, often neighbouring farmers' plots, and time. Initial experiments covered a wide range of factors but within three seasons farmers had identified similar areas of concern which coincided with those of experiment station researchers worldwide. Farmer experimentation and the role of the researcher in this methodology for technology development are discussed.

The effects of component population, rhizobial inoculation and applied nitrogen on the efficiency of phosphorus utilization in maize/mungbean intercropping were examined in two experiments. The efficiency of phosphorus absorption decreased by between 5 and 43% in maize and by between 31 and 58% in mungbean as a result of intercropping. The overall efficiency of phosphorus absorption in intercropping decreased with increasing nitrogen level but a parallel increase in overall conversion efficiency maintained a constant and large land equivalent ratio (LER) up to the highest level of nitrogen. The increase in LER over unity, however, was due primarily to the greater total absorption of phosphorus by maize and mungbean together in intercropping compared with that when they were grown in pure stands.

Restriction fragment length polymorphisms (RFLPs) are genetic markers based on cloned molecules of DNA. Using RFLPs, scientists can construct high density genetic linkage maps and locate economically important genes. Once a gene for a valuable trait has been mapped with RFLPs, desirable genotypes can be selected using RFLPs rather than scoring for the trait itself. This is important when a trait is recessive, difficult to score, or obscured by other characters. RFLPs are particularly valuable in mapping genes controlling complex polygenic characters, including those fundamental to crop improvement. The first RFLP maps were developed in crop species primarily important in temperate countries. Now RFLP maps for major tropical and sub-tropical crop species are also being developed.

On-farm trials were conducted for seven years in north-west Syria to investigate the use of land left fallow after a barley crop to grow feed legumes for small ruminants. Common vetch (Vicia sativa), forage peas (Pisum sativum) and common chickling (Lathyrus sativus), with and without applied phosphate fertilizer, were grown in rotation with barley and a barley-fallow rotation was used as a control.

Vetch was preferable for grazing whereas chickling yielded more grain and straw and had a higher harvest index. Both species showed substantial responses to phosphate fertilizer. Yields of subsequent barley crops were slightly lower after unfertilized feed legumes than after fallow but after fertilized feed legumes, barley yields were higher than after fallow. The milk production of ewes and liveweight gains of lambs grazing vetch, chickling and common land were similar.

The project showed how results from on-farm research can modify on-station research objectives and identify questions for further on-station research.

Field experiments were conducted on very alkaline calcareous soils to investigate the effects of two proprietary foliar micronutrient fertilizers, Wuxal Suspension Polymicro and Bayfolan 11-8-6, containing iron, zinc, manganese and boron, on the seed cotton yield of Gossypium barbadense and G. hirsutum. Two applications of foliar fertilizer, one at flowering and one three weeks after flowering, increased yield by 40–68%. The foliar fertilizers enhanced earliness and minimized nutrient depletion from the leaves, and thus increased reproductive efficiency.

Pot and field investigations were conducted to study the effects of vesicular-arbuscular mycorrhizal (VAM) inoculation and triple superphosphate fertilization on nodulation, dry matter yield and tissue nitrogen and phosphorus contents of Bradyrhizobium-inoculated soya-bean and lablab bean in the Sudan.

Inoculation of both crops with the VAM fungus Glomus mosseae in clay and sandy soils in pots increased nodulation, dry matter yield and tissue nitrogen and phosphorus contents more than triple superphosphate fertilizer, but even greater responses were obtained from G. mosseae combined with fertilizer. Crop responses in the two soils were similar, despite the large differences in soil physico-chemical properties.

In the field, inoculation of both crops with any of four VAM fungi enhanced nodulation, dry matter yield and plant nitrogen and phosphorus contents more than did triple superphosphate. Gigaspora margarita and Glomus mosseae were superior to Gigaspora calospora and Acaulospora species and resulted in more extensive root infection, especially in soyabean.

Three seasons of on-farm experimentation to develop potato production from botanical or true potato seed (TPS) under cool fertile conditions in West Java, Indonesia, are described. Twenty-three farmers experimented with two production systems: use of transplants, and use of seedling tubers produced in nursery beds. There was little yield difference between the systems, but an apparent progeny × system × season interaction was observed. All progenies were more resistant to late blight than the present cultivars grown from tubers. Appropriate matching of progeny and system gave seed of comparable quality with, but total yields slightly less than, certified imported seed of cv. Granola. Ware quality was slightly better than that of cv. Granola. Production costs were markedly less than for a tuber crop, making TPS ideal for small, resource-poor farmers.

On-farm diagnostic experiments were used to study constraints to nitrogen fertilizer use, define recommendation domains, and test fertilizer management techniques for two sorghum cropping systems in semi-arid tropical India. In the post-rainy season sorghum cropping system, nitrogen fertilizer use remains negligible because crop response is poor when sorghum is planted after the end of the monsoon, as is traditional. An earlier planting date would introduce high risks. In the dependable rainfall region, nitrogen fertilizer use on hybrid sorghum could be enhanced by tailoring recommendations to broad farmer-designated soil groups, and by educational programmes on the use of complex fertilizers. Additional research on fertilizer efficiencies on shallow Vertisols is needed. Diagnostic experiments proved a useful, relatively rapid and low cost method of research when complemented by farm management and on-station research data.

A two-year study was conducted at two locations in Niger to compare the response of five cowpea cultivars to two planting patterns and two sowing dates relative to the sowing date of pearl millet. All the cowpea cultivars took less time to mature when planted three weeks after millet than when sown one week after. Early-maturing cowpea cultivars had the smallest yield, whereas the yield of the local indeterminate cultivar was nearly double that of the other cultivars at both dates of sowing. Early planting of cowpea significantly depressed millet yield, but when cowpea was planted late millet yields approached those of the sole crop. The local cultivar depressed millet yields more than the other cowpea cultivars. Planting pattern had no significant effect on the yield of either cowpea or millet. Cowpea yield was reduced by more than 50% by two weeks delay in sowing, confirming that the sowing date of cowpea relative to millet is a critical factor in the Sahelian pearl millet/cowpea intercropping system. There were marked differences in the effect of cowpea cultivar on millet yield but these differences were consistent over sowing dates.

A mobile sampling method was used to measure the amount of photosynthetically active radiation (PAR) transmitted through the canopies of coconut palms in two long term field trials. PAR transmissions were measured at approximately monthly intervals for one year. Fractional interception of light was found to vary considerably between months, and to be associated with frond shedding caused by dry season water stress. Palms receiving potassium fertilizer intercepted more light and replaced leaf area lost during the dry season more rapidly than those receiving no potassium. Palms planted at a higher density intercepted more light and carried more fronds per palm, but intercepted less light per frond than those planted at a lower density. The ratio of moles of carbohydrate yield to moles of intercepted PAR quanta was calculated for six plots of hybrid palms, giving a dimensionless light efficiency index (LEI) comparable to harvest index. Mean LEI for all plots was 1.275 × 10−4 and was found to increase with increased light interception, results suggesting that there is a basal level of light interception below which no yield is produced. The potential for using LEI in agronomic experimentation on sole cropping and multicropping and as a criterion for genetic selection of coconut palms is discussed.

Cowpea is an important supplement to the staple diet of maize in Zimbabwe, but its productivity is poor beacuse of pest infestations. This paper reports the results of experiments conducted over two crop seasons on the feasibility of producing a grain crop of cowpea intercropped with cotton to take advantage of cotton's slow initial growth and of drift from insecticides applied to it. Intercropping reduced the yield of cotton by between 11.5 and 73% depending on the type of season and the treatments applied, but the reduction was compensated by the yield of cowpea. The yield reduction was greater in the drier first season and with simultaneous planting of the two crops. Cowpea consistently performed better as an intercrop than as a sole crop, mainly because of the protection it received from the insecticides applied to the cotton.

Yield stability in cassava requires genotypes that produce well under the variable moisture conditions encountered during the growth cycle. Plant characteristics related to yield stability were studied in two cassava clones subjected to 105 days of water stress in a field drainage lysimeter. Stress conditions commenced 117 days after planting, and the plants were allowed to recover at the end of the stress period for the rest of the growth cycle. Water stress restricted the growth of leaves and stems, but root yields were increased or remained unaffected. Leaf water potential varied little with stress, but gas exchange rates were about 75% those of the control throughout the stress period in both cultivars. Under stress, the plants partially closed their stomata and extracted deep soil moisture slowly. A high yield in both wet and stressed environments was associated with high mean LAI, better leaf retention, and greater partitioning of shoot biomass into leaf formation.

Seedling growth from sowing to emergence of two crust tolerant and two susceptible sorghum genotypes is described. Tolerant genotypes had longer mesocotyls with faster growth rates than the susceptible genotypes. The mechanism involved in crust tolerance appears to be that of avoidance by fast growth.

Beans and bananas were grown in sole crop and intercrop systems for three seasons in 1989 and 1990 to determine the effects of cropping system on diseases, insects and land-use efficiency and the roles of certain bean plant physiological traits on bean yield in the two systems.

Bean yield in the intercrop system was 52% of bean sole crop yield. Shading reduced yield, but stem starch levels, chlorophyll content and the chlorophyll a:b ratio were not related to seed yield in either system. Leaf area index was linearly related to yield in the intercrop system, suggesting that a higher plant density may result in higher yields. Nutrient concentration levels in the foliar tissues indicated that low potassium and high manganese availability constrained intercrop bean yield. Leaf manganese levels were higher in beans grown under bananas, reaching levels normally considered toxic for beans. Competition for other nutrients did not appear to affect bean yield. Apparently, interspecies competition for soil moisture did not much reduce bean yields as moisture levels did not differ with cropping system. Disease and insect pest levels were similar for the two bean production systems.

Banana yields were associated with potassium levels in the soil. Bananas appeared to be more competitive than beans in the intercrop system. The land equivalent ratio of the banana and bean intercropping system over the three seasons was 1.60.

Thirty on-farm trials were conducted in northwest Syria over a period of four seasons to examine the main effects and interaction of sowing date, sowing method, Rhizobium inoculation, phosphate application and weed control on chickpea (Cicer arietinum). Advancing the sowing date from spring or late winter to early winter resulted in substantial yield increases. The effects of Rhizobium inoculation were small and inconsistent. Chemical weed control significantly increased grain yield compared with the unweeded control treatment in winter-sown crops but was less effective than hand weeding. Phosphate application significantly increased yield in the first three seasons, the overall increase being 10%. Drilling chickpea seed gave an overall 10% increase in yield compared with the broadcast sowing commonly practised by farmers. There is no economic or environmental risk to the farmer in adopting the proposed practices either singly or in combination. However, the greatest gains in yield and net revenue are derived from a combination of early winter sowing, drilling, weeding and, where appropriate, phosphate application.

Sustainable crop production on Rwandan oxisols is limited by widespread soil acidity caused by high levels of exchangeable aluminium. This study was designed to test the effectiveness of an indigenous lime material in counteracting the acidity and enhancing crop yields. Lime application significantly raised pH, exchangeable calcium and effective cation exchange capacity, and reduced exchangeable aluminium and total acidity. Calcium was directly proportional to effective cation exchange capacity (r = 0.962**) and was inversely related to aluminium (r = −0.955**). Consequently, yields of wheat, beans and potatoes, which served as test crops, were significantly increased by liming. Lime at high rates (4–8 t ha−1) had a longer residual effect than at low rates (less than 2 t ha−1), suggesting frequent applications are needed when low lime rates are used. Simple regression analysis showed an increase in pH of 0.154 units and a decrease in exchangeable aluminium of 0.385 meq 100 g−1 for a tonne of lime applied.