Nikon F 35mm f2.8s

[Robert Camfield]

In comments from a few years back—2019 or so—Roland opined that the 6 element/6 group variant of Nikon’s 35mm f2.8s for the F was likely the best, optically. I have an unorganized collection of the more common pre-auto focus F lens, and exploring this claim has been on the agenda for some time. I got to it last week...the usual ad hock test including tripod and common subject (a street sign), distance, and lighting. So, yes, the 6-element version of the 35 2.8s (i.e., the early AI variant, which also sold as a K lens) appears to be the best, notably in the corners at all apertures. Interesting to me is the successor 5 element lens (late-AI and AIS): it’s pretty good, despite the simple optimal performance. I should mention also that, for resolution, the original 35mm f1.4s—in this case, an early production sample (#351676)—seem to be at least as good as the 2.8s, for apertures f2.8 – 8.0.   

[pluton]

This comports with my observations: The 35/1.4, once stopped down to f/2.8, is the best of all.  My old several-times-disassembled copies of the 35/2.8 K/Ai 'New Nikkor' have much field curvature at infinity, but seem to be optimised at a mid-distance, like maybe 2-3 meters. Pre CRC, Nikon seemed to often pick a medium distance to optimise the rendering for, instead of infinity.  Pre-CRC, this was probably a good idea for their reputation with pros and photojournalists who mostly shoot people doing things at intermediate distance.

[Robert Camfield]

Thanks for your comments. To clarify, the distance for these comparisons was about 30 meters...Nikon may well have optimized performance shorter distances for these early 35s, as you suggest. Generally speaking, ad hoc comparisons such as these tend to confirm the consensus view. Accordingly, the 24mm f2.8 AI/AIS appears to have noticeably higher resolution than the 24mm f2.0 for apertures 2.8-5.6, with performance of the f2.0 (in my case, an AI variant) approximating that of the 24 f2.8 by f8/f11.

[pluton]

Birna discovered, a few years ago, that the Ai-era 24/2.0 is very sensitive to the internal CRC mechanism being knocked out of proper alignment.  This physical vulnerability has led, over the decades, to reports and observations of some copies of the lens being very good and others being noticeably not as good. I think that the famous 28/2.8 AiS has this same vulnerability.

[Robert Camfield]

Yes, it seems that I recall Birna mentioning the issue...also, Richard Haw indicated, in a response to an earlier post, that the CRC mechanism of the 28 f2.8 AIS could be calibrated, suggesting variation across samples. My 24 f2.0 AI disappoints, while the 28 f2.8 AIS proves to be superb.