Your description suggests the document has acquired some of corruption. Corrupt documents can often be 'repaired' by inserting a new, empty, paragraph at the very end, copying everything except that new paragraph to a new document based on the same template (headers & footers may need to be copied separately), closing the old document and saving the new one over it.

Alternatively, it may be that the table underlying the graph has acquired some of corruption. Corrupt tables (which the above process won't repair) can often be 'repaired' by:
• converting the tables to text and back again;
• cutting & pasting them to another document that you save the document in RTF format, which you then close then re-open before copying them back to the source document; or
• saving the document in RTF format, closing the original document then opening it the RTF version, then saving that in the doc(x) format.
Do note that some forms of table corruption can only be repaired by the first method. I'm also not sure what effect this process might have on your graph.

Having said all that, I'm wondering why you don't do all this tracking & graphing in Excel. An appropriately-configured Excel worksheet would allow you to track the data continuously and produce graphs spanning any period of interest.

__________________
Cheers,
Paul Edstein
[Fmr MS MVP - Word]