When a log graph isn't quite as useful

 

I normally use log graphs to make curved graphs straight. The Covid data from the Office for National Statistics was impressively straight but it now appears to have peaked. I can't even say that the gradient of the curve is proportional to the R-number. It is a linear relationship and so linked but not directly proportional. So in general, the bigger the gradient, the bigger the R-number. This week's Covid case numbers for England go like this:

So the ONS is lagging the test data by a week.