An amusing gaffe in an ISO standard

As many of you are painfully aware, I have lately been writing a bit trying to demystify the intertwined matters of ISO speed ratings for digital cameras and exposure meter calibration. This has of course required me to review the various ISO standards involved. These aren't usually a source of entertainment value (nor tutorial value either). But in the ISO standard for exposure meters, ISO 2720, I ran into an inexplicable curiosity regarding the infamous meter calibration constants, K and C, I had seen before but glossed over. This time, I looked into it thoroughly, and uncovered an amusing, and inexcusable, gaffe.

It's hardly anything that really affects us, but I was so amused by it I thought I would write up the story in an article, "An Amusing Gaffe in the ISO Standard for Photographic Exposure Meters" (catchy title, wot?). It is available here:

http://doug.kerr.home.att.net/pumpkin/index.htm#ISO_Gaffe

On further reflection

I have recently been in an extended conversation with our colleague Helen Bach, who is very familiar with the topic of my article.

She points out that the fact that ISO 2720 takes the rather peculiar route of having different values of K for the standard exposure equations in the two forms (using the ISO speed in its arithmetic vs. logarithmic forms) cannot be demonstrated as being the result of an error in the algebraic work of the standards developers.

In fact, I now recognize that it might have resulted (for example) from a well-intentioned desire to keep the equation for K2 (like that of K1) free from any pesky constants.

While one might question the judgment of the developers in embracing (for whatever reason) this clumsy result, it cannot be demonstrated (at this point in time) to have resulted from an actual error in their work. (I will of course still harbor that suspicion!)

In any case, given my recently-enlightened view of this, I have decided to withdraw my article for the moment. Whether the topic warrants a new version of it remains to be seen.

In any case, thanks to Helen for taking the time to work with me on this matter.

Best regards,

Doug

A little more discussion

I thought I would expand a little more on my new outlook on the curiosity in ISO 2720.

The scaling the standards body adopted for the calibration constant "K" (considering first its appearance in the standard exposure equation in the form in which the ISO speed is expressed in its arithmetic form) is arbitrary. The scaling they chose has the nice feature that no constant is required in the equation, giving it a lovely simplicity. And there would be no motive to select any other scaling, since "K" had no pre-existing definition or usage: it is only a creature of this calibration equation. (I make no claim that this was the rationale for the scaling they chose.)

Now, while we might think that it would have been suave for the standards body to hold the same scaling for K when it appeared in the standard exposure equation in its other form (in which ISO speed is expressed in its logarithmic form), they really had no obligation to do so. That scaling is equally arbitrary, and they were free to choose a different (arbitrary) scaling there.

What was their motive for actually doing so? I have no idea. But it is interesting to note that by being willing to have two different scalings for "K" in the two forms of the equation, the second form of the equation, like the first, is also free of any annoying constants. (I make no claim that this was their motive.)

Note, to the credit of the developers, having taken that course, they actually gave different symbols for the two calibration constants: K1 for the one intended for use in the first form of the exposure equation, and K2 for the one intended for use in the second form. That's a good thing, since for any given "calibration strategy", the two K's would have different numeric values (differing by the factor 1.259, as a matter of fact!).

In any event, whatever the motivation of the developers, since the scaling of "K" is arbitrary anyway, and nothing says that K2 can't have a different arbitrary scaling from K1, we cannot say that the mathematical train in the standard, or the relationship between the two forms of the exposure equation, reveals any "error", or demonstrates that the outcome was a result of any error (any more than we could if someone writing a model specification for pool tables decided to state the width in yards and the length in paces - dumb, maybe, but not the result of any "error").

Now, as to whether it was "suave" for the developers to have done so, my opinion is that it certainly wasn't. I think the resulting structure is clumsy and can confuse the user. For example, if a manufacturer harks to the invitation of the standard to mark the chosen value of "K" on the meter case, they would actually have to state both K1 and K2 (since almost any meter of that era allows entry of the ISO speed in either the arithmetic or logarithmic form)!

Fortunately, since (under modern ISO standards) ISO sensitivity is no longer expressed in the logarithmic form, the matter is moot.

Best regards,

Doug