1st of all there is too much noise in the baseline. In the spectra taken in the positive mode the peak at m/z, 638 may be your M+(molecular ion peak). But in the spectra taken in negative mode, it's very difficult to identify molecular ion peak due to so much noise in the baseline. So, i think this is not informative. You can check column contamination to eliminate the noise. You can also change the column programming to get a better spectrum.

Mr. Shade,

For structural anaysis there are needed CID-MSn (n=2 or 3).

Hi, I agree with Dr, Ivanova, it is nice if you can have ESI MS MS; by using adducts you can predict molecular ion, and by ESI  MS MS data you predict the chemical structure of your compound (s) by using some free on line MS data bases (Metfrag, Metlin, MassBank, Metfusion etc). If  M+H 638.20 it should be have M+Na etc.

http://fiehnlab.ucdavis.edu/staff/kind/Metabolomics/MS-Adduct-Calculator/

Dear Dakota:

  You are very fortunate in having these spectra. Next, you need to make arrangements  to submit them to the U.S. National Bureau of Standards.  Contact your IT people or whatever other agency at your institution can guide you to this U;.S. government site. MS spectra can be matched to an enormous database and give you a positive ID for the compound in  the spot. 

Good luck. 

Dr. Sonenshine

@Noel Davies Actually the solvent from TLC was butanol/ammonium hydroxide (8:2) and that's where the sample was purified from. I understand I tell the actual compound without more data, but I needed to be able to explain this spectra. Thank you very much

How about the negative spectra?

I also found a table of common peaks from other sources and the 638 was attributed to silica from injector and a few other peaks were matched with that.