I think we are converging 🙂
The nr. of elements (actually the number of surfaces) was the degree of freedom designers had to correct for a number of aberrations, not just center resolution. Generally, the fewer the number of air/glass surfaces, the more transmission and contrast you get.
Historically, lenses back in the days were designed like this: designers picked a proven design, and then assembled a system of linear equations for each lens surface (Pen 38/1.8: 11, Canon 35/2: 12), which optimized at least the Seidel aberrations at image center and corner, for different focus distances, via manual ray-tracing, and by making conscious trade-offs. Then they solved this system of equations. When you read early lens patents of simple designs, you can follow this process. In the 1950s in Japan, rooms full of people would solve equation systems manually and in parallel (via Gauss elimination, for instance). Later, this was done by computers (the earliest I know of is the DR Summicron for which Mandler used a Zuse Z5 in the mid 1950s; for being "only" f2, the early Summicrons are extraordinarily complex).
You will not find a middle-corrected Planar design as in the Canon for 50mm equiv. lenses in any format. These were specifically used for moderate wide angles (as in the early 35mm Summilux) to optimize distortion and field resolution at high speeds (f2 and faster, on FF). Mid 1960s forward, for most f1.8 and above normal lenses, designers used 6 element Planar (10 surfaces) or Ultron designs (as in your 38/1.8); these were extended by additional rear elements for "fast" normal lenses, as in, say, the Pen Zuiko 42/1.2.
Roland.