What I recall is that the PK>MK switch on the 3800/3880 does not need as much ink because a little contamination of MK ink with PK will not be visible on matte media. The other way around and a minimum of MK ink left would show gloss difference on gloss media. So here it is PK>MK 1.6 ML versus MK>PK 4 ML.
When the x900 series were announced I wondered about the changes on that system. You might know, is the PK/MK switch near the head on the 3800/3880 so the black ink tubes (2) are not affected or is it near the cartridges so also the black tube (1) has to be refreshed with either black? I would assume the first solution is used but it will make the ink carriage + tubes assembly heavier on a relative small printer. For the x900 I can not see another solution than a switch near the heads, it would take way more ink on the long tube and way more time for refreshing if the switch is near the cartridges. My initial thought when they claimed less ML used on the x900 models was this difference in switch systems. But if both printer types have the same switch near the head then I find it strange that a x900 would use less ink on the switch cycle, channels and ink buffers near the head should have larger dimensions for a wider format with 360 nozzles per inch per ink channel (speed). It could be that your ANC story reveals another explanation; splitting the ink waste in two and using different terms for the wastes. A gain in less waste by a redesign of the switch near the head (bringing it even closer to the black nozzles) is of course also possible but why did the 3880 not get the same treatment then? Ink sales + printer service is the main part of the profit for printer manufacturers and there may be some differences in that approach between the 17" and 24"> models, ink per ML already differs. So I am not sure what actually happens but find 1 ML for half a cycle a low number that is not telling all.
Met vriendelijke groet, Ernst
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2014 update, 700+ inkjet media white spectral plots
As the saying goes, your mileage vary. Approximately two years ago I did the ink swap test on my 9900, both directions(PK>MK and MK>PK). I suppose the difference from the original post to my procedure is that I used an actual "ink" swap, not water. However, I doubt that would or should have any effects on the final outcome. My procedure was to use an empty waste tank with a small container to catch the ink and then do the measurement. Over a period of a month I repeated this procedure three times and averaged the results to take into consideration any possible measurement variables. At that time I was initiating the ink swap in standard operating mode, so of course there was also the accompanying cleaning cycle to consider. My averaged readings were as follows:
PK > MK = 3ml
Mk > PK = 5ml
Total ink dump during swap = 8ml
All prices are CDN Dollar. 350ml Cart = $.34/ml
Total cost per two way ink swap = $.34x8ml = $2.72
With the CDN Dollar at its current rate I dare not rework the figures for a current swap.
Also, for some time I have been doing the K ink swap in Service Mode to avoid the auto clean cycle that always follows the swap in Standard Mode. I assume that by removing that factor from the equation the total cost per two way swap would be somewhat less.
The CL1 vs Std. cleaning ink usage was a surprise to me as well. However, it's a moot point, since I have always found that the CL1 pairs cleaning is more efficient than the same cleaning procedure in Standard operating mode. In Standard mode it's often a matter of losing more nozzles following the first clean cycle, as opposed to Service Mode where a CL1 clean will almost always clear the nozzle on the first try. Just my experience.
Gary