Hi Bob,
My first choice was, of course, the capacitors. Unfortunately, I had forgotten that I had some leaded capacitors less than "4 pF" from another project. So, the crux of my question was whether those resistors shown in the Microchip schematic could be be used to pull it slightly lower.
Then, I remembered that I had some mica capacitors down to 1.0 ± 0.5 pF and ceramics a bit higher . They are relatively huge but solved the problem. In another thread, it is asked what is on our "to order" list. I have added some 3.9 pF ceramic, COG capacitors, as my choices were 3.3 pF and 4.7 pF. The latter was too high. My capacitor meter is a DE5000 LCR, which is generally quite satisfactory, but fails in sorting such low value capacitances.
My main concern was linearity about the target frequency. From 436 kHz to 564 kHz, the response is absolutely linear (given 0.1 kHz resolution), so I am not particularly worried about that small offset now.
Right now, I am on a bit of a tangent to get a faster 32-bit x 24-or 32-bit division routine.
John