Tower-fly-by has always been one of the most precise methods of altimeter calibration. As part of an exercise preparation at ITPS, I flew on a number of TFB test points in the Czechoslovakian Aero L-29 Dolphin to verify the limited numbers found in the AFM and check Mil-Spec compliance.
While the concept of the Flight Test Technique is pretty straightforward, the data reduction and analysis required for Mil-Spec compliance is not. Is there more than one Mil-Spec? Which one should be chosen? What do they require? Are they currently cancelled?
In short, the MIL-I-5072-1 covers all types of pitot static tube operated instrument systems while MIL-I-6115A deals with instrument systems operated by a pitot tube and a flush static port, as is recommended by USN TPS. The quantitative numeric side of the tolerances described in the above two specs is exactly the same. USAF however, seems to have taken those specs a step further through MIL-P-26292C(USAF), which puts more focus on high Mach number and supersonic region and defines the tolerances based on parameters derived from dimensional analysis (ΔP/qc) rather than absolute tolerances in knots and ft. For data presentation purposes, all specs require analysis of the errors to sea level conditions vs. indicated speed parameters, which again is not a single step process.
While all specs are currently cancelled without replacement, probably due to the many new and non-traditional air data systems and the air data computers used, they remain a good guidance especially for older airplanes.
Unsurprisingly, as the figure below shows, the L-29 failed to meet the MIL-P-26292C(USAF):