Why fitting without thought can be a terrible idea

From the figure caption: "The dashed line is a quadratic polynomial (given by y = -0.516 + 1.08x - 0.23x^2) which gave the best fit to the points."

David W. Roubik, Science 201 , Sept 1978, pg. 1030.

Dr. Pitre believes the data looks more like a duck!

Illustrating how we may have more than one minimum in the chi-squared in a non-linear fit