Your protein might indeed be aggregating or forming multimers such as dimers, trimers, etc... (I haven't worked with globulins, so I'm not sure if they form multimers or not). You should be able to calculate from your elution profile if the first peak to emerge is in the void volume (aggregates) or if it corresponds to the size of a dimer, trimer, etc... Since you mention that both peaks present themselves as completely clear (no turbidity)and their SDS profiles are also the same, I'm inclined to think that you have some form of multimer in the first peak, as you would probably see more contaminant bands or even a smear in the SDS profile of the first peak if it contained aggregates.

As you also mention that your prep isn't very clean after gel filtration, there's also the possibility that one or more of the contaminating proteins are interacting with your protein of interest.

Since your protein doesn't carry an affinity tag, I would recommend you add an ion exchange step to your purification protocol as this might remove at least some if not all of your contaminants.

Also, I would use DTT instead of betamercaptoethanol, as it is a stronger reducing agent and is active for longer.

Thank you Antonio Ariza  for answering my question. The peaks are not in void volume. No matter how i do, i get same two peak with same elution profile and SDS-PAGE profile. Yes i guess there could be one or more contaminating proteins which are interacting with my protein of interest.

As the protein is only soluble in presence of salt. Because of insolubility in absence and low salt concentration iam unable to do ion exchange step.

Yes tried using DTT instead of mercaptoethanol, but no change. The SDS-PAGE profile in presence and absence of reducing agent is same.

The proteins I work with are also very salt sensitive as they have a very high pI (>9). They don't like ion exchange but it's possible as long as I keep the protein very dilute (

We have used ammonium sulfate precipitation followed by zonal isoelectric precipitation, the latter to separate different types of globulins. It's a method first published by Shutov and Vaintraub. This paper by Casey (1979) Biochem J 177: 509-520, that applies the method and is more accessible (in English) http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1186401/pdf/biochemj00470-0131.pdf

In essence, the column is packed with gel flitration media that will exclude most proteins and is equilibrated to an acidic pH (e.g. pH 4.7) in which the proteins are not soluble. The proteins are in a buffer in which the proteins are well dissolved (e.g. pH 8). When the protein solution is charged onto the column, the proteins precipitate forming a white ring at the top of the column. The buffer dripping into the column is switched to the one that the proteins are dissolved in, gradually increasing the pH of the eluate. As the pH increases, the proteins are gradually eluted, depending on their different solubilities. The buffers have salt (phosphate, NaCl) and reducing agents throughout the procedure. You might need to tweak the zonal isoelectric precipitation conditions to suit the pI of the seed proteins in the species that you are studying. There may be some overlap of the different globulins so it is advisable to monitor fractions by SDS-PAGE in addition to monitoring by absorbance at 280 nm. For soybeans, we have found this to work very well for isolating 11S globulins. Purification of the 7S globulins required additional steps   http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1056068/pdf/plntphys00605-0077.pdf 

how can we reduce the formation of multimers as i am working on hfq which is a hexameric chaperone protein and is his tagged. On SDS PAGE gel i can see a large aggregated band at the top of the well plus a monomeric band  exactly at its molecular weight.also the absorbance at 280 nm for my eluted protein is too high that is 1.3