Mention polynomial calibration in readme for changes from v1.X.

README.md

@@ -35,11 +35,12 @@ Work to do includes:
 Migration from < v2.0
 ---------------------
 
-For most users the primary difference from v1.X will be that the data to predict
-and fit is now specified as a list of `prospect.observation.Observation()`
-instances, instead of a single 'obs' dictionary.  There is a new convenience
-method to convert from the old 'obs' dictionary format to the new specification.
-This can be used with existing scripts as follows:
+For many users the primary difference from v1.X will be that the data to predict
+and fit a model to is now specified as a list of
+`prospect.observation.Observation()` instances, instead of a single 'obs'
+dictionary.  There is a new convenience method to convert from the old 'obs'
+dictionary format to the new specification. This can be used with existing
+scripts as follows:
 
 ```py
 # old build_obs function giving a dictionary
@@ -64,6 +65,10 @@ from prospect.fitting import fit_model
 output = fit_model(observations, model, sps, **config)
 ```
 
+Another change is that spectral response functions (i.e. calibration vectors) are now handled by specialized sub-classes of these `Observation` classes.  See `spectra`_ for details.
+
+.. _spectra: docs/spectra.rst
+
 Finally, the output chain or samples is now stored as a structured array, where
 each row corresponds to a sample, and each column is a parameter (possibly
 multidimensional).  Additional information (such as sample weights, likelihoods,

doc/spectra.rst

@@ -39,13 +39,17 @@ over the polynomial coefficients (this allows you to place priors on the
 accuracy of the spectrophotometric calibration). Or you can just take the
 spectrum as perfectly calibrated.
 
-Particular treatments can be implemented using different mix-in classes, e.g.
+Particular treatments can be implemented using different mix-in classes for the
+:py:class:`Spectrum` observational data, e.g. for optimization of a 5th order
+polynomial calibration vector at each likelihood call, use
+:py:class:`PolyOptCal` as follows
 
 .. code-block:: python
 
         from prospect.observation import Spectrum, PolyOptCal
-        class PolySpectrum(PolyOptCal, Spectrum):
+        class PolySpectrum(PolyOptCal, Spectrum):  # order matters
             pass
+
         spec = PolySpectrum(wavelength=np.linspace(3000, 5000, N),
                             flux=np.zeros(N),
                             uncertainty=np.ones(N),