This is of no relevance to the OP but I found myself wondering what the best approach might be to the problem of averaging top and bottom pairs of an arbitrary sized set of numbers. There are many possibilities but one I liked was to define a Named formula 'extremes' to refer to

**= IF( {0,1}, LARGE(numbers, {1;2}), SMALL(numbers, {1;2}) )**

The result is then given by

**= AVERAGE(extremes)**

Since we appear to have a group of contributors that are comfortable with array formulas, here are a few thoughts on array formulas that have made an impression on me, together with my recollection of from whom I have heard an authoritative statement.

All calculation within Excel is performed by an array calculation engine - Joe McDaid

There is no separate calculation code that operates with scalars, they are simply 1x1 arrays.

References to arrays within the cell are reduced to single values by implicit intersection - Joe McDaid

CSE is no more that an instruction to omit this additional step.

All Names are references to array formulas - Daniel Ferry

They may be thought of as named ranges or named constants but they are still just formulas.