I agree, but what could have changed?
bq. the induction system is just a pipe with some valves at the end of it
Well, there is a bit more to it, isn’t there? 
The likely culprit, in my opinion remains ignition, or induction.
I don’t think it is surprising considering the design of components involved here.
As you have demonstrated, you have found a big improvent and that is so much better than the other way around.
Continue to monitor the GAMI spread over time and it is very likely you’ll find another change in future.
Have you ever tried an inflight mag check, deep LOP at a reasonably low power setting? That’ll show up any diference between l/h and r/h mag and demonstrate how good the mag(s) are really set, no matter what shop carried out the work on your behalf.
LOP is your friend finding these issues. ROP unfortunately hides most issues.
VFR (low level, 23" all the time, slightly LOP, and the flights are usually short anyway) or IFR (and one runs out of MP at all normal Eurocontrol levels
VFR and IFR have no influence on leaning regime.
Try running your engine WOT (wide open throttle) all the time from take off.
Why close it? Why add a restriction to the air pump?
Lean to control power output, your rings will thank you for it.
That means LOP below 7000 feet when flying your normally aspirated engine.
Above 7000 feet only RPM will give you more power at peak or slightly lean. Your CHT and oil temp needs to be consulted.
Very high up you might find slightly rich acceptable, but you are operating at 50% or less power.
Biggest problem is the reduction in cooling mass in the air.
I think you are ready for a turbo, don’t you think? 
A few years later, I have re-done the GAMI test. This is actually on a replacement engine (from Jan 2017) on which the injectors were re-matched. So three years later I did this new test
The spread from the first to peak to the last to peak is 0.3 USG/hr which the GAMI chap says is perfect.
Having said that, I think the 6th polynomial curve fitting in Excel 2007 is dodgy. The peak on #6 is not well placed. There is also this basic issue: the EDM700 stores the data in quite a low resolution.
Funny how #2 is so much coller than the others, and it always was.
An old thread… @complex-pilot, @commander and @cheshunt may still be around?
The absolute EGT value is of no consequence, only the relative value referenced to peak. The aim is to get all cylinders peaking at about the same fuelflow.
I remain puzzled regarding the physics which can result is such lower temperatures while the engine still appears to deliver smooth power.
The explanation must be in the airflow.
Obviously all engine temps must track the ambient air temp*. And #1 and #2 should be the coolest because they are at the front. There is an explanation for #1 and #2 being different from each other; it is because the two are staggered, with #1 right at the front and partially blocking the hole in the cowling so that, evidently, #2 gets a better airflow around it.
And sure enough here we see a delta of about 15F between #1 and #2 CHTs, and the same delta between #1 and #2 EGTs
which is reassuring
* with the exception of the oil, whose temp is thermostatically controlled, and therefore parts of the engine which see a significant heat transfer to/from the oil will also not track the OAT.
CHTs are mostly related to airflow and baffling inside the cowling, EGTs are also influenced by the location of the probe in the exhaust tubes, distance from the exhaust valve, depth inside the tube etc.
If the engine is running smoothly at peak EGT or LOP, that means all cylinders are developing the same power, that is the aim of GAMIjectors.
Here’s a question for those with Lycos and Gamijectors :
According to Gami TCM/Conti engines are more likely to profit from Gamijectors due to their design when compared to Lycomings.
I intend to do the “Gami lean test” and send them the data for our Commander’s Lyco IO-540 6-cylinder for their evaluation whether it would be a good investment.
This said, I would be curious to have feedback of actual experience from those pilots who fly Lycos with Gamijectors. How much is the gph gain when flying LOP ? Is the engine running measurably smoother or is it just a vague feeling ? Overall, is the switch to Gamis worth it ? Again, from what I read there is no doubt about the significant improvement for Contis so this question is meant for Lycomings only.
How much is the gph gain when flying LOP ?
You mean when running LOP rather than ROP? Well, that would depend how far ROP one was previously running.
Also, when you analyze this, you want to be sure you are drawing the same HP LOP than ROP. (Of course, just taking a ROP power setting and then leaning the mixture to somewhere LOP will not achieve that; it will leave you with much lower power setting).
One good way of doing that is noting the speed (flying ROP), then going LOP and adjusting power to get the same exact airspeed. That will tell you the real GPH savings of running LOP.
In a 260hp IO-540, I would expect typical savings to be about 2 GPH (again, at the same HP). But as I said, that would depend on how many degrees ROP you were operating in the first place.
In the famouts “Mixture Magic” column by John Deakin, there was a nice photo of a twin’s engine instruments, with one engine runnung ROP and the other LOP (but both at the same horsepower), which showed a difference of 3.2 GPH:
Is the engine running measurably smoother or is it just a vague feeling?
Again you mean LOP vs. ROP? No, an engine will always be less smooth LOP than ROP, since the power curve is steeper on the LOP side. Hence, when LOP you will always have higher diiferences in power outputs from the various cylinders (which is what creates roughness).
I actually meant before and after the GAMIjector installation when flying LOP.
My consideration/goal is extending the range.
“when LOP you will always have higher diiferences in power outputs from the various cylinders (which is what creates roughness).”
Then I must have misunderstood something. I thought roughness would come from the difference between the cylinders being “fed” unevenly, the so-called Gami-spread, i.e. some cylinders running LOP, others at peak EGT and some possibly ROP. I understood Gamijectors were supposed to narrow this spread and hence take away some of the “roughness”.
