<<I see you are an energy economist. This explains a great deal. >>
Actually it explains nothing.
<<Your print usage needs to be put into context as according to your own methodology a 'print' measures 54 square inches. Therefore, 1300 standard prints in four months equates to about 75 near full-bleed A3+ prints per month. Nothing wrong in that, but a print house would have a much larger throughput and they would be the most interested in the economics you have laid out. Also, you admit that you cannot log all type (i) cleanings and none of the type (ii) cleanings. >>
I can log all the type (i) cleanings, there are few type (ii) cleanings and they happen with both machines. With the 4000 I may have missed several type (i) cleanings. They consume 4 ml each. No big deal as a fraction of the total.
<<If there was a significant difference between your throughput with the 4000 against the 4800, these cleanings may have had a significant effect on your data.>>
They don’t have much effect on the data relating to ink used for printing alone, because just from recollection of what the machines do while printing, I am certain that type (ii) cleanings are considerably less important than type (i) and type (iii) cleanings. To the extent the 4800 needs less cleaning altogether, there have probably been proportionaly fewer type (ii) cleanings, so that is part of the overall performance improvement.
<<Next, you say you print with profiles. Do you make them yourself or do you get them from elsewhere? Have you considered the possibility that your typical profile for the 4800 is simply a 'better' i.e. more economical profile than that you would have used typically for the 4000? A different profile might yield a radically different result for either printer. >>
Everyone must print with profiles, as you know. 99.5% of all the prints I made with both machines use Epson profiles for Enhanced Matte paper. If they've improved their Enhanced Matte profile in respect of ink consumption for the 4800 that may be one validating reason for the improved performance of the 4800.
<<Also, your own data indicates quite significant variability in ink usage per session (a spread of about threefold). Yes, I know that in 77 out of 110 of your sessions, the ink usage is close to 1 ml per SP, but even in these 77, there is variability of up to 20%. >>
There is variability from the 4800 also. To date, the minimum ML/SP has been 0.54 and the maximum 1.00. Once we get into the kind of print numbers I've done, and visually see with each session how little the overall average wiggles around as more and more data gets added to it, one develops confidence that this variability is not randomly impacting conclusions in either case.
<<By your own admission, you only print on to matte media, so your data is not applicable to someone like me who prints mostly on glossy (RC) media. >>
Agreed. As I said, I'm not a testing lab. I'm a guy who prints his own pictures - with some seriousness, and I wrote the article to give readers on this website whatever benefit there is to sharing my experience.
<<Finally, you have changed from the 4000 to the 4800. Can you honestly tell me that you would have published the data or have been as enthusiastic about the results if you had found the opposite. >>
Having presented my first article on the 4000 to Luminous-Landscape for publication, the follow-up story for the successive printer model is a natural, regardless of the results. Of course if nothing had changed the story would be less interesting, but findings of either substantial improvement or substantial deterioration make it more useful to readers whichever way the cookie crumbles.
<<Sorry, but I do find sometimes that there is too much discussion of pseudo-science or pseudo-stats here. >>
No need for apologies. These are not pseudo-stats. The data is carefully and correctly logged session by session, subject by subject. The exercise is scientific to the extent that the methodology consists of conventional addition, subtraction, multiplication and division embedded in a fairly straightforward Excel spread-sheet and completely repeatable by anyone who wants to use it.
<<Give me passion for quality anyday, hence my interest in the qualitative differences between the printers. I note that you (and MR) do not think there is a huge amount of difference in quality, >>
Well Drew, cost-accounting and appreciating quality are two separate aspects of all this, and like you I very much appreciate the latter to the extent that I put a great deal of time and effort into it. From comparisons I have seen of 4000/4800 prints on non-matte media (i.e. what you use) there is a quite noticeable difference in quality - if bronzing and gloss differential matters to you. I've never been all that impressed with those issues because if you look at the print from one angle you see the problems and from another angle you don't. It's a bit elusive, but many serious photographers find brozing and/or GD totally unacceptable. This is a matter of taste and standards, which vary legitimately from person to person. I agree that you need to see these things for yourself and come to your own conclusions based on what you see, what matters to you and hence what you know.