Thank you very much Glenn, much appreciated. My matrix is larger than the extract I posted (the times extend to 2200 rather than upto 0930 as I posted) so I'll need to have a think about how to scale the solution to capture the entire dataset. However, you've given me a new avenue to explore - I was bogged down in my own SUMPRODUCT formulae that were stubbornly refusing to work.