The Remove references window is shown above.  To use it you need to specify:

• Whether references are 'downwards' or 'upwards'.  A 'downwards' reference -- such as in the example -- is one in which the reference data have lower values than the sample data, and an 'upwards' reference is one in which the reference data have higher values than sample data.

• The reference index value.  If you have specified a 'downwards' reference, the program assumes all data with values less than the index value are to be considered reference data (vice versa if you specified an 'upwards' reference).  In the example, the index value is set at 0.022. 
          You can also use the cursor to select a reference index cutoff, using the 'Set reference cutoff with cursor' button. Move the cursor until the indicated line is acceptable, and then click the mouse.

• The number of extra cases. Reference readings are seldom instantaneous events; usually there is a gradual change -- over a fairly constant number of samples in a particular data channel -- between 'true' sample and 'true' reference values.  This period of change needs to be excluded from reference removal calculations.  The value you specify as 'extra cases' is the number of excluded samples on either side of points identified using the reference index value -- in this case, 12 samples on the left and 15 on the right.

• Whether reference data are interpolated between initial and final values, or replaced with a fixed value.  In interpolation, the computer fills in the reference gap by 'drawing' a line between the initial and final values.  This is essentially the same operation as can be done manually using the 'linearizing mode' option in the spike removal window. 

Reference removal is performed and the new values are displayed in the plot area when you click the 'Fix references' button.  You can then undo the results, or exit the function.