Polynomial fitting generates erratic plot

Hello, I have a distance/time curve. The polynomial fit is erratic. A cubic function fits the data well however this does not start at zero as it should but the error is acceptable. Is it better to use fplot to replot the data and calculate the optimum values or use a different approach?

Once the plot has been created I will take its second differential and from this calculate the minimum.

Gavin.

[dpb]

Hard to judge w/o even a picture but what about interpolating spline instead? If the data are noisy then smoothing may be appropriate, however. But, we "no nuthnk!" useful in that regard from which to judge from the information given...