In eqn. 17.2.18 of the 3rd edition, the coefficient 0.01 should multiply the first term in the denominator, not the second. The code Dopr853.h that implements the routine is correct - this is just a typo in the book.

Thanks to Freddie for pointing this out.