Thursday 24 November 2011

Well, that changes everything. Perhaps.

I have heard it said that sometimes the most exciting moments in science are when you look at something and say “that’s odd!”. I had one of those moments today. I have been working on the paper from my post-doc research this week, and the main focus of my work has been the figures. I now have very good, clearly labelled BSE images of every sample analysed and am starting to reference them in the text of the paper. The process consists of saying things like “Garnet ranges in texture from thick rims overgrowing seeds to a spray of very small grains” and then adding parenthetical notes behind each type pointing to figures that show each sort. In the process I also listed those samples which have no garnet at all, with reference to their photos.

This is where the “that’s odd” moment came in. For one of the samples my table of results shows that it has garnet, and that I have five good analyses of it. However, the photo of that sample shows that there is no garnet at all, unless you count the seeds of a vastly different garnet composition that were included in each experiment to give the new garnet an easy place to start growing. Clearly something is wrong here, so I checked the data file and determined that the five garnet analyses we have for this sample were done by my predecessor in this project years before I arrived. In my own records I also have three analyses of garnet seeds from this sample. Comparing his data with my own shows that the garnet analyses he did are clearly *not* the seeds—for this sample the seeds are grossular garnet (the calcium end member) but the garnet he analysed are iron-rich with only a little Mg and Ca. Sorting all of the garnet data for all of the samples shows that his analyses for this particular sample fall nicely in the middle of the pack in terms of composition.

So what is going on? I can think of a couple of logical possibilities off the top of my head. One (kind of obvious one) is that I somehow analysed the wrong sample myself and the photo I have is not the sample I think it is. Another is that I did analyse the correct sample but there is something wrong with his data. A part of me hopes that it is the latter—not just because of vanity and the hopes that I did everything perfectly, but more because it would clear up a problem that has been bothering me with my results since I first obtained the data.

I ran my experiments at three different temperatures: 600, 625, and 650 C. For the two higher temps garnet was present for every pressure I tried. For 600 C, on the other hand, I found no garnet at all in any of the metapelite runs, but it was present in 3 of the 4 metagreywacke runs. Now, had those three been adjacent to one another I wouldn’t have complained. But they weren’t. Instead I had garnet present at 2.2 GPa, garnet absent at 2.5 GPa, and present again at 2.65 and 2.8 GPa. This makes it difficult to draw pretty diagrams. How much nicer they look when a phase is absent for one half and present for the other half of the diagram.

So now I sit and eagerly await a reply from my erstwhile boss to the email I just sent him. Depending on how he replies I will have work to do changing the write up to match the new situation—do I get to delete the analyses of garnet for this sample that he did, and write the paper with a story that makes sense to me, or will I need to beg him to find the real sample and put it into the microprobe and take new BSE images and do an analysis or three to confirm that the composition my predecessor saw all those years ago are correct, or…?

1 comment:

Alison said...

That is indeed very odd, I hope that you will post what your resulting decision is... Actually I hope that your erstwile boss will run more data, just 'cos I am that kind of curious....