June 18th

Today I tried the forcing the calibration line through zero for the data set I have been working with, and I found that it was no different than the weighted linear fit. I also tried constructing an actual concentration versus reported intensity graph myself, and seeing how to best fit that curve. I found that a polynomial fit worked the best, and my result was exactly what the computer had done for its fit. Next, I designed and prepared a new calibration series, one that is equally spread over the range of concentrations we are working with. We also made a few changes to the testing method the machine uses, changing the amounts of time it is permitted to read each sample. I think that these two changes should help eliminate a lot of the systematic error we have been seeing.