Various difficulties and issues in developing MAKEE echelle extraction procedures --------------------------------------------- (1) Pixelation problems. With a curved order the fractional amounts that pixels near the boundaries of the source varies. If not done correctly this produces a repeating sawtooth pattern in your spectrum. The solution is to use a profile model to appropriately compute the contribution of a fractional pixel. (2) Trace problems. Jumping the gaps. An polynomial fit to the trace centroids is needed to extrapolate over gaps in the spectrum when following the trace. (3) Low count problems. This is a subtle problem related to the failure of Poisson statistics at low count levels. The weighting of pixels by their invariance variance fails. To fix this a profile model is used to compute the correct variance.