This is on ieee.org. The article mentioned this[1] other test by the Chicago Tribune. This isn't purely an advertisement though it serves that purpose a bit.
It's disputed because they only tested 2 iPhones.
> There are reasons to take the results with a grain of salt, however. McCaughey clarified that Penumbra supplied RF Exposure Labs with one iPhone 7 and one iPhone 11 Pro for the tests—phones the company had purchased off the shelf. He attributed not testing more phones to the cost of purchasing multiple iPhones
This is also important:
> More notably, when the FCC conducted a follow-up investigation after the Tribune published its story, the agency did not find evidence that any of the phones exceeded SAR limits. That said, while the Tribune and Penumbra both used off-the-shelf phones, the FCC largely tested phones supplied by the manufacturers, including Apple.
It raises the question whether Apple and others supplied the FCC with phones that are different from what they sell.
[1] https://www.chicagotribune.com/investigations/ct-cell-phone-...
That weasel word “largely” isn’t helpful. They either only tested supplied phones, which may raise suspicions; or they also tested phones sourced from other channels, however insignificantly, suggesting problems with the third-party tests, or they would have noticed the anomaly. “Largely” ostensibly points to the latter, but it could also mean “we don’t know”.
What weirds me out is that they’ll send them to these third party labs and repeatedly test some small number (possibly one) of devices until it’s right up against what’s allowed. That feels less than scientific and probably wrong although I’m not sure what I’d change.
Such behavior has already been shown by automobile manufacturers during the "diesel-gate" incidents.
Remember that this is fundamentally a legal/social thing that involves engineering/science and not the other way around.
Before the tin-foil crowd calls regulators stupid, I'd like to point out that in the absence of cheating this allows hardware engineers to do a better job & design a system that is robust to software shenanigans. That's a nice thing. Maybe we cannot have nice things, but if so, let's prove it rather than presuppose it.
I'd estimate the likelihood of cheating by looking to see if people are choosing phones on the basis of EIRP the way they choose cars based on MPG. I don't see that happening, so I estimate that the likelihood of cheating is low, and would vote to keep the cooperative model around.
“Double the energy limit” sounds a lot, but for a lot of these labs that’s inside of the error range of their equipment.
> Penumbra was conducting the test, which also included testing an iPhone 7, to study its Alara phone cases, which the company says are designed to reduce RF exposure in a person