Almost all of our sensors use a two point linear calibration. This means that a specific voltage produces a specific value of whatever you are measuring. For example, a change in voltage coming from the pH probe of 0.1 V corresponds to a change in pH of "one."
|pH Sensor Voltage||pH
The physiology sensors behave a little differently. The ScienceWorkshop software has to find heart rate from the EKG voltage the same way you would; by determining how long between QRS peaks and extrapolating out to one minute in order to find the number of beats per minute. There is an algorithm in ScienceWorkshop which steps through the EKG voltage trace and tries to determine which peaks "count" (which ones are actually the QRS's and not the T's). The algorithm requires five "real" data points in order to determine heart rate from the EKG voltage.
The extremely high heart rates you see at the beginning of your experiment or when the baseline voltage changes are caused when the algorithm doesn't have five real peaks to operate on. When the algorithm finds its first peak, it tries to calculate heart rate using one "real" peak value and four other random numbers. When it has two peaks, it uses two "real" values and three random numbers etc. In the case of the changing baseline voltage, the algorithm finds the peaks incorrectly until the baseline voltage stabilizes and there are five "real" peaks.
To remove the bogus data from statistics calculations, highlight the data you want by clicking, holding and dragging to draw a box around only the data you want to use in your analysis. Only the data in the highlighted region will be used in the statistical calculations. In order to turn the highlighting off, click anywhere in the Graph display.