FIGURE 1. Schematic of classification of farming systems in the Eastern District of Dominica. PRESENTATION OF THE CLASSIFICATION > 10 acres "Small Planters" I High Banana <10 acres --Agricultural Income activity alone Progressive FarmersII Agricultural activity as secondary income generating activity Part Time Farmers IV Land is not a Limiting factor (> 5a) Ilia Not much banana ( / land (2 acres) Land is a limiting factor (< 5a) IIIb Important Agricultural / Activity Labour is a A limiting factor IIIc Low Banana Income Needs a limiting Factor Unsettled farmers Va Limited Agricultural Activity '%Labour as a limiting factor Aged Married Couple Vb Labour as a limiting factor Woman alone Vc Land as a limiting factor (< 5a) Agricultural set of constraints that must be identified before any technical or economical development proposal can be implemented for him. As one cannot study the case of each individual farmer, it is ex- pected that they may be grouped in several homogenous types with minimal variations inside the classes, so that a proposal help- ing a farmer of this type to overcome some of the constraints he meets will probably be of interest to the other farmers of the same type. Several approaches may be taken to identify and study these types. Theoretically, one can take a structural or a functional ap- proach. In the structural approach, one studies a sample of farmers' structural parameters of agricultural activities (for exam- ple, size of tenure, work-force or available capital), and the distribution of these parameters. In the functional approach, one concentrates on the agricultural activities themselves (type of pro- ductions and crops, level of intensification, technical "routes" followed, etc.) These two approaches are complementary, since the basic assumption of F.S.R. is that there is a regular correspondence be- tween function and structure, based on the "rational assumption" cited above. In both cases, the types of farming systems identified are only hypothetical ones, and need to be checked by more detailed studies (monitoring) on several farmers of each type. In the case of the TRD project, however, a very light approach which may be described as empiric, was preferred. This was due to the limited manpower available, and the concern of both na- tional authorities and the local team not to spend too much time on research, and the desire to start the training and development of activities as soon as possible. A first set of 40 interviews was carried out in January 1982, in one "average" village of the area (Morne Jaune), collecting both structural and functional data. Using "data analysis" methodology (but with manual means), a first stratification of six classes ranging from "investor/job integrated in the farm," to "small farmer without land," was obtained. The stratification was mainly based on the past evolution of the farm and the objectives of the farmer. From July to August 1982, six farmers belonging to the dif- ferent classes were monitored, through a daily or weekly data col- lection, by students attached to the project. From January to December, 1984, four farmers were monitored by the project team. A second classification was proposed in July 1984, based on the first results of these monitoring. This classification takes more structural data, acreage, other regular sources of income, and labor force into consideration, and is more detailed, including five classes, and ten subclasses. The observations realized during Extension activities are also integrated into this classification, which remains, however, an empiric and hypothetical one. Use of This Classification - Further Investigation Needed As indicated above, this classification has to be considered as a hypothesis until more detailed studies, including a statistical analysis, are conducted. Nevertheless, it can be used as a training tool for Extension Officers and other officers involved in agricultural activities, and as a working tool for starting develop- ment activities with the farmers. As a training tool, it points out the fact that there is not one average farmer representative of other farmers of Dominica or of the district, and not even one average small farmer or medium farmer. Therefore, there is not one development action which is valid for all farmers at the same time, but several different pro- posals should be adapted to different types of farmers. This fact is not unknown to the Extension Officers who face it every day in their work, but it needs to be theorized and analyzed in order for them to draw practical implications from it, and be able to return this "field information" to higher levels of the Agricultural Department. A practical example that has been studied with the Extension Officers is one of pasture development. For many years Extension services have been proposing that farmers interested in livestock establish pastures by fencing an area with wire and planting im- proved grasses ("Pangola Grass"). After many years of efforts, it can be seen that most of the pastures established in this way, though still fenced, have been overgrazed and have gone back to a "bush savanah" vegetation. The first conclusion could be simply that the technical proposal was not adapted to farmers' condi- tions, but when one analyzes which type of farmer did what, one finds that two different types were involved in this pasture establishment. 1. Type I, petit planteurs or progressive farmers, have land and capital, but no labour available. For these farmers, fencing a pasture is a way to decrease the work needed in livestock maintenance and still try to intensify animal pro- duction and increase the livestock capital. All this leads to a high probability of pasture destruction. 2. Type IIIb, diversified farmers who have limited land, but labour available. Their problem is to increase their income (to meet family requirements) by producing marketable production. If they raise livestock on pasture, they will pro- bably try to increase the production by increasing the size of their stock. This may lead to overgrazing as in above case. These farmers may be interested in an alternative pro- PROCEEDINGS of the CARIBBEAN FOOD CROPS SOCIETY-VOL. XX Labourer Vd 112