You sound very energetic and sincere.  I dont know if you want advice from a hydrologist with some basic soil knowledge, but I would suggest to adjust your strategy for perhaps greater gain and targetting of result.  When soil scientists classify soils in US, there is the taxonomic classification and then the descriptive classification of each soil series. Soils series are identified and mapped by soil scientists based on physical, chemical, hydrology and sometimes plant indicators to facilitate this mapping.  I can tell from your effective question that you are relying on farmers, villagers or citizens to tell you which soils are different and perhaps delineate.  I do not detect if you have knowledge if you are dealing with 4-6 soil types or perhaps many more if you are looking at varying areas.  I would first suggest you take some time to enlist help if needed to identify and classify soils to soils series, then stratify your sampling based on soil series.  You can still do your added analysis in sampling a given number of sites within each soil series and probably 10 in each soil series would be a good start, look at results and see how much variability exists, if it is high, perhaps more samples would be needed.  I have a friend who is a certified soil classifier and have worked with him for many years.  I believe he would tell you the same, work within the existing soil classification framework, rather than require such extensive testing to classify soils by fertility.  At some point, even with all these fertility tests, one must still make a decision on whether the land should be farmed, forest, produce wildlife habitat, scenery, or be developed for housing, industry or something else.  The US soil scientists have a category that they call prime farm land and perhaps that is something you are seeking.  They classify certain soil series as prime farm land due to not only its nutrients, but landscape position, water holding, physical properties, etc. so it deserves to be identified in a special category.  And there is a soil series called Leon, which is a spodisol, wetland some years with nutrient imbalance and physical limit with iron concretion issues and not very good for tree growth.  If you have an area with plastic soils, they will be difficult to manage with heavy equipment when wet, or soils frequently flooding, another problem that may limit their use, or for some crops might even be a benefit.  In addition, and no less important, if you get the soils classified to series, there may be other information already available on these soil series that you can compare your data to.  I would at least see if information on soil series is available and the criteria they used to map them apply to some or all of the soil areas you are seeking to classify.  During my career, soil scientists have remapped soils several times to meet new criteria and map new soil series that have been identified.  But even the old soil series mapping efforts have information that might be useful.  If you do not correlate to the old system of mapping, the new system may be costly or demanding in effort to be applied on landscape scales.  If you are not aware of or cannot find any information on soil series mapping through USDA -NRCS, let me know and I will see what I can find an example for you.  The soil scientist that I mentioned knew how to read plants, topographic position, soil characteristics, etc. to put soils in a series where most management questions can be answered without a lot of soil testing.  An excellent soil classifier should have a great career with all the work wanted and may also provide advice on whether soils will perc for septic systems, whether soils need certain nutrients to be most productive, whether soils have operational limits, or whether soils are wetlands with specific critical habitats, etc.   

If you stratify your soils into soil series from all sites, see what information is already available, run 10 samples per soil series in stratify random sampling, you should have plenty of information to decide if a new or refinement to classification is justified and desired.

I agree that it would be useful to begin with existing soil maps, if available. Of course it is not opposite to use soil types indicated by farmers. If zones from soil maps are not the same, as indicated by farmers, for example, one soil series on the map comprises three "farmer's" zones, I would collect samples separately from each of these zones, at least 5 samples (it is my opinion, the absolute minimum would be 2 samples, but, from the statistical point of view, more samples - better).

If you see part of field, which seems clearly different from the rest, you might omit it (if it is small) or sample separately (if it have greater area).

I am afraid, that 24 soil samples is insufficient, it would be good to represent one greater field or village, but not the region. If the farmers of one village indicate 4 different soil, you need to collect 20 samples or more (in this village, 5 or more samples for each soil).

Another issue to resolve is if you collect separate sample from one point, or rather composite sample from determined area. I think the sampling per point is better to characterize soil variability, sampling per area is better for elaboration of fertilizer recommendations. However, in both cases you need to have several samples (even composite samples) per one soil type indicated by farmers, soil series presented on soil map or area, which is clearly different from the remaining part of the field/village. When you form one composite soil sample, it is recommended to make it from about 15 up to 40 single samples taken from one, apparently homogenuous area.

Another issue is the depth - generally, many agronomists limit sampling to topsoil or arable layer (0-20 or 30 cm) to characterize soil (chemical) fertility status. However, the crop yields (not only of perennial, but also annual crops) depend not only on chemical fertility on soil, but even more - on water conditions (shortage or excess of water). To assess it, you should sample soil at least to a depth of about 1 m - not necessary to all analyses, but at least to determine soil texture.

The last question is even more complicated and we have not sufficient data - it depend on field, crop, soil etc. For example, if none of the measured properties is yield limiting, the variability accepted might great. I think you should first detect the main yield limiting factors. To do it, you need not only soil data, but also yields and frequently meteorological data (especially water balance - i. e. precipitations and potential evapotranspiration). I would start from simple regression analysis, for example I would analyse relationships yield vs soil separately for dry, average and wet year. And than other methods, for example multiple regression...

Thank you Sir, Both of the responses are very elaborate. I have realigned my approach as per suggestions. To begin with let me first understand the soil classification data that already exist. I will do preliminary study and then write back with queries. If any information for Maharastra, India is found pls indicate to me. 

These comments have enough insights for me to begin reading work with a fresh perspective. Thanks again.

 Mr.Oak ,you have raised a good question.I appreciate the elaborate comments of Dr.Hansen and Dr .Stepien on the subject.Soil series based information for Maharastra is available with National Bureau of Soil Survey and Land use Planning(NBSS&LUP),Nagpur.They have brought out state-wise publications on soils/soil series of different states.They have also reports on studies carried out on farmers' fields on soil series basis.You may visit their website for more details or personally contact and purchase their publications and maps.You may see the correspondence or difference when you superimpose soil series information on farmers'perception in a particular village.In particular  soil series ,based on resources you have,the farmers' resources(irrigation,manure fertilizer they can use) and the level of management (low medium and high) by farmers, you can sample 25 to 100 surface samples.

Is there an solid procedure to begin the identification, classification and mapping of soils of a precise area?

I guess there are many, but could you cite some serious and replicable ones?

Assuming that the area is totally unknown (no existing soil maps, no help from the farmers).