This is the answer to Mahanty -

You just can compare de eluate of soil A and soil B for the same solvent. Otherwise the results Has not significance.

You can to the extraction with both solvents separatly and after mix the solvents or use one after other and compare....

Dear Celeste, I calculate a soil property based on C extraction value and D extraction values; i.e. C-D is the thing I wish to compare between A & B soil type. MY question is not about methodology, my point is how can I statistically compare the data set for the two soil types?

Still more clarification of definitions is needed Biswanath. You mention 'soil types' and  'subgroups', and the latter could have several different meanings, e.g. a subgroup is either a set of random samples within a type, or a subgroup is a recognisable subtype.  Statistical methodology is not my field.   

Subgroups, I mean replicate. Sorry for confusion. So I have two soil types each type of soil is extracted with either solvent C or solvent D (replicate). I have some hypothesis that the C-D reading is of important property. And for the said property, I want to compare the two soil A, B?

I have had a biometrician do exactly that for some of my data and it is not an unusual statistical question given that we migth be able to experimentally observe two or more 'fractions' in soils.  Are you attempting to relate the difference to some characteristic such as better growth?  That is how I would expect the modelling to be done.  I wish you luck with the stats

You are exactly understanding my concern... I am looking for some fraction from differential extraction, and I am comparing this difference across two soil...

But don't understand how to process the raw data in this difference normal distribution. Thanks for your reply

You can use a geostatistical treatment, using krigging, for example.

But if you use graphic representation is easier to understand the similarity and  the opposite behaviour, for each element you can use in a graphic representing both soils , in x axis put the concentration

You need to confirm either to compare the characteristics of the two soil types or the methods you use to characterize the soils or both.

Dr. Mahanty ,As I understand simple Paired  t- test may serve your purpose.Please go through the procedure in any agricultural statistics book.