Robin, the presenter in the YouTube link, covered a lot of material in depth ad so deserves detailed comment.
First, he never said he was "scraping the level in" although that was the initial plan. He stated his grinding job was probably closer than he could scrape and pointed to 30 micro-inches linearity thwartships and 20 longitudinally on his instruments. I agree but with a reservation I discuss below. My experiments with several combinations of transfer and contrast media (Prussian blue and red lead are still the best over-all in my book) suggest 40 - 50 millionths is about the lower limit of readable scraping indications. If he scraped the surface in to his best accuracy he might have satisfied sacred tradition but there might be a small sacrifice in flatness.
He went on to mention that his work with a scraper was unnecessary but was included for appearances sake. You notice there was hardly a ripple in the indications of the gage amp readings. But, we all have to admit the pattern and flash of the surface he finished was most attractive.
Something is nagging at me about the way the linearity of the level's reference face was proved. The level was dragged past the gage head registering it was parallel to that point on the plate. If the level was stationary and the gage head stand was moved over the surface plate to scan the level's reference face, the level would be proved by the flatness of the surface plate. If the level was reversed and a second set of readings taken with the gage head stand following the same track the error could be calculated and the true error of the level's reference face determined independent of the accuracy of the plate. I think. My head is spinning.
I hope all paid attention to the presenter's demonstrations of the effect of heat on dimensional and geometric accuracy, lead drag, etc..
He also mentioned Airy points Vs Bessel points and the uses of each, a distinction I've been unaware possibly because the distinction occurs in the third decimal place of the constant..
Airy Points and Bessel Points - Engineering Notes
Anyone note his use of precision diamond ground stones for burring scraped surfaces and dressing mag chucks? I tried that once and after a few uses the stone accumulated a $hitload of pins and scratched finishes .I wonder what he did differently?
I would liked for Robin to have discussed the accuracy of his granite flat on which millionths were hanging in the balance.
Another point Robin missed: master precision levels are factory scraped deliberately concave by 0.0002 to 0.0005" chord per foot. The theory goes this eliminates random teeter when the level is placed on a flat surface and if supported symmetrically when spanning gaps ensures a steadier end bearing. The concave is small but necessary I'm told. I've scraped several levels flat but never noticed any uncertainty of readings. I suspect the effect of a concave scraped level reference face is one of those distinctions without a difference beloved by marketers - but I hesitate to assert this notion as fully credible.
I thought the Robrenz video most informative. I learned several subtle points. I guess there's hope for me: 75 in a few weeks and still learning stuff/