Allow me to explain the waveform thing better, perhaps more clearly. The outgoing pulse from the generator is a simple wave (one vertical peak). The outgoing pulse passes by the oscilloscope (EKGs use a modified kind of these, they display the heartbeat on an amplitude-versus-time graph), and gets read as it passes. The outgoing pulse travels through a wire, reflects off air or water or meat, and the reflected pulse returns to the oscilloscope (the trip separates the reflection from the outgoing pulse). The reflected pulse is a complex wave (with wierd shaps and little molehills around it), because it contains information about what it reflected off of.
The oscilloscope does 100X10^6 samples per second, but it's just barley enough to read this wave. The result is that the data I work with is about 200 samples long, and about 8 of those data points will describe most of a wave. Because there are so few points on any given reflected wave, there's way too much opportunity for variation on one temperature/concentration setup.
So I've told the program to do an averaging of 1000 waves--so in my final data set each 8 points of interest are an average of a 1000.
Because my 'waveform' is really the scientific equivalent of some kid's Connect-the-Dots drawing (or maybe a Cubist trying to paint a tree), it's going to be practically impossible to regress linearly, with polynomials, or even non-linearly because any equation acurately describing the shape as a whole would be too bulky to work with.
My idea to deal with this is to take those 200 samples for one wave, and compare each point individually with another wave at a different concentration. So I take 10 waves for ten concentrations, and I take 1 position (out of 200 positions) and regress how it changes over the 10 concentrations. What I get is a much more simple relationship between amplitude and concentration.
It is this simple relationship I want to regress--to find an equation describing one point on the wave. With a spreadsheet program I can do this 200 times, and be able to describe the waveform as a whole using 200 equations.