I once did some research into a meter accurate enough for me to calibrate the EG&G myself. It was somewhere between 4 to 5 thousand. And a new EG&G sensitometer around that time was also in the neighborhood of 4 thousand.
I had this screaming argument with someone from Kodak back in the 90s. He was negating my findings because I had used an intermittent sensitometer instead of an non-intermittent sensitometer. I'm sure for Kodak, who use densitometers that read to 4 decimal places, who wrap their dip and dunk machines in thermal blankets, who use carbon step tablets and revolving step wedges, my set-up was lacking. To me, it produced pretty reliable and very repeatable results.