Thank you for publishing this

[John Hollenberg]

WIR testing procedures are said to be scientific and rigorous. There is no ISO standard yet for print longevity, but one reads now and again that WIR is looked upon as the closest substitute. They consult to museums and other institutions for whom longevity is a professional concern, and have built up a tremendous reputation in this field. We are not talking about flattering results, we're talking about dependable results produced in a transparent and repeatable manner (cf. e.g. Harald Johnson, "Mastering Digital Printing" Second Edition, and page 153 where the issue of recirprocity failure in permanence testing is also noted). 

Mark,

Can you clarify whether the "Wilhelm years" includes a correction for reciprocity failure or not?  Joseph Holmes told me that it didn't.  That is an important consideration, because reciprocity failure was stated to average around 2.5.  If the results aren't corrected, 100 Wilhelm years would mean 40 "real" years.  Quite a difference.

--John

[Haraldo]

Mark, Can you clarify whether the "Wilhelm years" includes a correction for reciprocity failure or not?  Joseph Holmes told me that it didn't.  That is an important consideration, because reciprocity failure was stated to average around 2.5.  If the results aren't corrected, 100 Wilhelm years would mean 40 "real" years.  Quite a difference. --John

I'm not Mark, but since Mark quotes me (page 153 of MPD2, with Mark's usuall attention to detail!), I can clarify that "Wilhelm Years" does not account for recripocity failure. If you go to the latest WIR ratings, which happen to be for the Z3100,
http://www.wilhelm-research.com/hp/Z3100.html
and you go to footnote #2, you will find the statement: "... High-intensity light-fading reciprocity failure in these tests are assumed to be zero...."

Is this is a problem? Could be. Why? Because it's possible that contributing failure factors may not reveal themselves until after a long period of time. But high-intensity accelerated testing is by definition, short term. How does one "correct" for reciprocity failure in accelerated testing? I'm not sure, but I was originally told by Henry Wilhelm that he was also doing long-term "real world" tests to validate the short-term extrapolations (which is what the WIR numbers are: they are extrapolations of print permanence based on specific dislay-life calculations). Is this still happening? I don't know. (Realistically, most digital printing products would be off the market if this were carried out in real-world terms.)

The real issue for me here is not so much that reciprocity failure is unaccounted for, but rather the total focus by everyone on the year specific numbers. I just change the years to "points." 100 points is better than 40 points on the WIR scale. Is it 2.5 times better. Probably not.

There are also other issues at play here, and there are some people working on different ways of approaching the issue of print permanence testing. I will report on them as they come up for me. Soon, in fact.

Harald Johnson
author, "Mastering Digital Printing, Second Edition"
DP&I.com (http://www.dpandi.com)
digital printing and imaging consultant

[John Hollenberg]

Harald,

Thanks.  I don't know where Joseph Holmes got the "reciprocity factor" as reported from 2 to 8 and probably averaging around 2.5.  I think the important thing is that 200 Wilhelm years is probably significantly better than 100 Wilhelm years.  Having an independent standard by which to compare the ink/paper combinations is the crucial thing.

By the way, by my calculation one year of exposure to light at 35,000 lux as Wilhelm does translates into about 155 Wilhelm years (factor of 2 there to account for the fact that his "years" assume 12 hours per day exposure of the prints to light, while the test presumably goes on 24 hours per day).  Is this correct?  If so, I would certainly think Canon should have some pretty significant data by now.  That's what has me "concerned".

Any other info you have on the intricacies of testing that I am not familiar with would be appreciated.

Thanks.

--John

[Mark D Segal]

Harald, thanks for coming in with that reply - you did a better job than I would have.

Regarding the intracacies of the testing John is asking about, I think a very good place to start is reading all the footnotes to a Wilhelm test - any one of them will do. I realize it is like reading the fine print on a mortgage loan, but nonetheless very educational. As well, Wilhelm has posted on his website quite a few scholarly papers in which he and colleagues describe both in theory and practice heaps of stuff about permanence testing.

[Haraldo]

One more before shutting down...

Harald, Thanks.  I don't know where Joseph Holmes got the "reciprocity factor" as reported from 2 to 8 and probably averaging around 2.5. 

Interesting, because I wrote about a "reciprocity factor of 2-4" years ago. As I recall, I got this directly from Henry Wilhelm at that time.

I think the important thing is that 200 Wilhelm years is probably significantly better than 100 Wilhelm years.  Having an independent standard by which to compare the ink/paper combinations is the crucial thing.

Yes. Being able to make relative comparisons has always been the main benefit of WIR numbers, in my view. But the WIR standard or testing protocol is based on certain criteria, and those criteria, while covering a lot of ground, do not take some things into account.

By the way, by my calculation one year of exposure to light at 35,000 lux as Wilhelm does translates into about 155 Wilhelm years (factor of 2 there to account for the fact that his "years" assume 12 hours per day exposure of the prints to light, while the test presumably goes on 24 hours per day).  Is this correct? 

What WIR does is expose the samples to a continuous bombardment of light until a failure point is reached, and then it extrapolates that to a "reference display condition" of 450 lux for 12 hours per day, which ends up being "Wilhelm Years." 450 lux and 12 hours per day is the reference for Wilhelm, or the "average user" experience that everything is based on. Kodak uses 120 lux.

If so, I would certainly think Canon should have some pretty significant data by now.  That's what has me "concerned".

Yes, this does seem a bit odd, doesn't it, especially in context with all the other printer models that WIR has posted results for. I have no concrete knowledge of why this is.

Any other info you have on the intricacies of testing that I am not familiar with would be appreciated.

All I can say is that ASTM (Mark Gottsegen and his Artists' Material subcommittee) has thrown its hat in the ring with a completely different approach -- see www.amien.org to contact Mark. And there is another player, too, which I will comment on very soon (on my website: http://www.dpandi.com). Again, a very different approach. Things are about to get interesting for print permanence.

Harald

[Ray]

Ray - no - its's very much right on track. WIR testing procedures are said to be scientific and rigorous. [a href=\"index.php?act=findpost&pid=109789\"][{POST_SNAPBACK}][/a]

Mark,
You've misunderstood me. The previous threads have been implying that maybe Canon have paid Wilhelm for results which are not being published because they are not flattering, not because they may not be accurate. As far as I know Wilhelm has a reputation for sound methodology. Lyson must have been furious when Wilhelm first published test results on their inks and papers several years ago. Presumably Lyson did not have any contractual agreement which gave them a legal right to stop the publication.

[madmanchan]

I think that's off track. Good results for publication in a scientific journal are accurate results based on sound methodology. Here we are talking about 'flattering' results.

Hi Ray,

I agree with your remark about how science journals work. However, my point still stands that if results are bad (e.g., because the technology being evaluated isn't quite there yet), then the authors won't publish them. If you look at bio, physics, and computer science journals, you'll see articles about methods and techniques that work, not method and techniques that don't work. In other words, you won't pick up a journal and find a bunch of articles that say, "We tried method XYZ and it didn't work because ..."  And even in published papers, you will often find that authors (deliberately) try to show the most promising test cases where the method works, as opposed to test cases where the method fails (i.e., failure modes).

How is this relevant? Well, naturally a company would prefer to be able to advertise WIR data for its papers that have good numbers as opposed to WIR data with bad numbers. You'll notice that not all of Epson's papers are consistently listed in the WIR data ...

[Mark D Segal]

Mark,
You've misunderstood me. The previous threads have been implying that maybe Canon have paid Wilhelm for results which are not being published because they are not flattering, not because they may not be accurate. [a href=\"index.php?act=findpost&pid=109847\"][{POST_SNAPBACK}][/a]

Thanks for clarifying. I think you, Eric and I are all saying the same thing. If Canon is not allowing publication it could mean they have reasons why they don't want the data out in the public domain. One thing Canon could do as a confidence-builder is to allow one of their reps to explain to us why there is no published WIR data for the IPF5000.