Although your question is somewhat cryptic, this may pertain to the fact that some studies have shown that the leucosomes in migmatites could potentially be veins of felsic melt introduced into the rock, rather than melt that has been produced in situ?

Hi There,

As Marlina pointed out your question is a little vague. My guess is that you are asking about reintegration of melt with residue to calculate the original composition of the rock that melted. 

Whilst this may be possible in some cases i think the likely scenario is that in many cases partial melting is open system. This means leucosomes represent residual cumulates that are left behind after the loss of some (in may cases unknown) volume of granitic melt that has now migrated to higher levels in the crust. 

As a result a mass balance approach will still not result in calculating the original bulk composition of the unmelted protolith. There is a JMG special issue on crustal melting that covers a large amount of current migmatite research (link below). As well as recent books/reviews e.g. Mike Brown, Ed Sawyer, there is also a lot of active research with modelling of open systems e.g. Chris Yakymchuk, Fawna Korhonen etc

http://onlinelibrary.wiley.com/journal/10.1111/(ISSN)1525-1314/homepage/virtual_special_issue_on_crustal_melting.htm

Dear Mr. Jafari,

have a closer look at the classical book published by

Mehnert, Karl Richard (1971). Migmatites and the origin of granitic rocks, Developments in Petrology. Elsevier

Mass balance calculation requires a statistical treatment and a precise definition of homogenous units as to size and shape and how they are interrelated with each other. Apart from the different views which still exist on how migmatites form; it is a very ambitious project involving different disciplines. I don´t want to discourage you but I am outspoken and rather pessimistic if it is possible to consider all the side effects to the extent necessary and it is worth the time and trouble to take such an approach.

Good luck!

Harald G. Dill