Some information on my testing, and what it means.
You have to recognize that manufacturers specify a default voltage lumen output, which is ONLY accurate if they test their bulb in an "integrating sphere" ("I.S.")--which measures the total output of the bulb from all directions. This is referred to as "bulb lumens."
However, not all manufacturers that list their default voltage bulb lumens have actually had the value independently verified in an I.S., because they are quite expensive.
Then the next thing to recognize is extrapolating higher lumens based on a Welch Allyn re-rating formula (on their website URL) as voltage and current changes....is not valid for other bulb types. It even becomes invalid for WA pushed higher than 25-30% of default voltage. Specifically, notice the values of this 1185 bulb for voltage and current in the URL. By manually inserting different voltage/amperage values in the URL, you can force the WA website to use a set of formulas to predict bulb lumen output at various inputs.
What you see on my destructive testing charts, under the "Predicted Lumens" column is based entirely on this re-rating formula from WA's website URL, once their formulas were figured out and put into a spreadsheet by CPF member AWR. In contrast, my Lux readings are ACTUAL tested results taken by light meter on 2 separate bulbs, mostly done on two separate nights. If the 2nd bulb test results did not correlate within a few lux of the 1st bulb tested (which was only the case with two bulbs), a 3rd bulb was sacrificed, and the two closest readings were then used.
This lack of objective data from bulb to bulb is why I did the destructive bulb testing. You can read my setup description in the initial post, and why I measured Lux at 1 Meter. I saw this testing as akin to Silverfox independently testing all the batteries at various Amp discharge loads.
I'm not even sure that you can use my Lux reading compared to manufacturer's default voltage bulb lumens, and extrapolate bulb lumens at raised voltages...in part because it assumes default voltage bulb lumens that the manufacturer listed are correct.
For example if you look at the 5761 bulb, Philips says at 6V it puts out 765 BL. My measurement gave 94 Lux at 6V.
So theoretically, if you wanted to know what an accurate bulb lumen reading would be at 7V, you could take my tested value of 143 Lux. Then you should be able to solve for x in this equation based upon the default "known" comparisons:
94 Lux 143 Lux
------ = ------ (cross multiply)
765 BL x BL
94x = 109,395 (solve for x)
x = 109,395/94
x = 1164 bulb lumens
Is 1164 BL more accurate than the WA formula predicting 1312 BL for a Philips bulb? Honestly, I don't know.
I do know however that an actual (repeated) Lux test reading of 143 Lux is a brighter reading than any of the Lux readings I got testing the WA 1185 bulb here. I also know that the Philips 5761 bulb is 52% brighter at 7V than at 6V....which is more reliable than the WA formula predicting it is 71.5% brighter.