The nine bar question
Nine bar came from the size of a lever spring, with no documented testing behind it. So how much does it really matter?
Lance Hedrick's video “The 9 Bar Fallacy: Why It Was Never Ideal” argues that nine bar — the pressure nearly every espresso machine is built around — was never backed by any documented testing, and that you're usually better off brewing lower. It's a good, provocative claim, and, unlike most coffee arguments, it's actually testable.
Most espresso machines only brew one way: a pump pushes water at whatever pressure a spring or a valve is set to, and it barely changes from the first drop to the last. A pressure profiling machine (such as our DE1 model) is different — it can change it as the shot runs, and it logs every shot. After watching Lance's video, I did three things: I checked his claims, I lined them up against Jonathan Gagné's writing — the physicist Hedrick himself leans on, and a fixture in our community — and I looked at what our community's logged shots actually do, thanks to visualizer.coffee integrating with our software to cloud-store people's espressos.
tldr: the three of us mostly agree, and the real story is more interesting than the video's title. Nine bar isn't magic, and more pressure isn't free. Where we part ways is on how low you should go, and on whether you can taste the difference at all.
Nine bar is a historical inheritance from the spring size of 1940s lever machines, not a number anyone arrived at through documented testing. And flow through the puck rises in the low-to-mid range and then plateaus — so the exact pressure in the working range matters less than how evenly the water moves through the puck.
— how we'd sum up the question Lance Hedrick's video, “The 9 Bar Fallacy,” raises (our words, not his)
Hedrick's argument has four parts. First, the history: pressurised espresso started with Achille Gaggia's spring lever in the late 1940s, which happened to make 8–10 bar and, with it, crema (different makers wound different springs — anywhere from about seven to eleven bar — so even the number was set by the spring's size, not by testing). FAEMA's E61 (1961) copied that pressure with a pump. Nobody tested their way to nine; it's just the number the hardware made. Second, that the old claim of flow “peaking” near nine bar is a misreading — the real figure in the literature is closer to five. Third, that between about seven and ten bar the exact pressure can be described as “a red herring.” And fourth, that a tasting panel can't reliably tell shots less than three bar apart. His conclusion: stop worrying about nine bar, and try brewing lower.
On the history he's right, with a footnote: crema had to be sold, exactly as he says — Gaggia had to market the “scum” floating on top before people wanted it. On the “five bar, not nine” point he's also right, and I can pin the source: the 1993 Petracco & Suggi Liverani result everyone cites actually put the departure from linear flow at ~5–7 bar (quoted in Corrochano 2015), not nine. The other two claims — how low is best, and whether you can taste it — are where Gagné and our own data have something to add.
Jonathan Gagné is the astrophysicist behind the Coffee ad Astra blog and two books on coffee physics. Hedrick leans on his work; so do I. He has written the clearest explanation of why pressure and flow don't behave the way you'd expect. Darcy's law — the tidy rule that flow is proportional to pressure — assumes, in his words, that “the medium of percolation is fixed, and immovable,” and for espresso that “is wrong in many ways.” As the pressure climbs, the puck “compresses and becomes shallower,” fines break loose and move, and the pores between the particles shrink. The result, from the numbers DE1 owners have gathered over the years, is that flow tracks roughly the square root of the pressure — so doubling the pressure comes nowhere near doubling the flow. He's blunt about where the tidy simulations mislead:
“It treats the bed of coffee as a fixed and immovable object… This is why their simulation tells us that 'the fastest flow is always best,' which may have you want to apply 150 bars of pressure on your pour over. If you did this however, you'd find that your coffee bed would quickly erode and channel pretty badly.” Jonathan Gagné, Coffee ad Astra
Note how additional pressure causes a reduction of flow around 9.5 bar. My hunch is that 9 bar was picked as the standard because it was the highest pressure possible before secondary puck compression starts to happen around 9.5 bar. I have no evidence of this, but it fits the facts — and gives more respect to espresso's founders than an arbitrary “follow what's been done before.” To be clear, that’s my own hypothesis, not the argument of the 2026 paper Hedrick leans on: it models the puck as poroelastic and describes the high-pressure side as flow saturation, not a hard secondary-compression ceiling at 9.5 bar.
An honest digression, on thin data. How strongly this secondary compression shows up — if at all — seems to depend heavily on the grinder and the coffee. The idea got its first real push when James Hoffmann challenged me on it years ago, so I asked people to run tests; John Weiss was one who did, and found flow dropping as pressure climbed past about 9.5 bar. Lance's hunch is that the coffee explains it: Weiss brews darker and favours a bimodal grind, which — Lance suspects — may let the bed compress earlier. His own testing points the other way: with light roasts and ultra-low fines on an EG1, he saw the onset of any secondary compaction move to higher pressure. This is an early trend on limited data, not a settled result — exactly the kind of question thousands of logged shots could finally answer, which is the crowd-sourced follow-up Lance floated to his own audience.
Gagné also identifies a hard ceiling. Above about ten bar, he writes, the pores start to close up and the shot chokes — the “secondary puck compression” he credits to me and other DE1 users. And he points out something the pressure knob hides: fines migration gets worse at higher pressure, because the water moves faster between the grains and drags more of them downstream to clog the bed. More pressure, in other words, actively works against you in a way Darcy's tidy line never shows.
New this year, a paper put numbers on all of this. Waszkiewicz et al., “Under pressure: poroelastic regulation of flow in espresso brewing” (Physics of Fluids, 2026), modelled the puck as a deformable medium and measured steady-state flow at 11 pressures from 1 to 12 bar. That's the data in Figure 1: flow rises to about 4.5 bar and then flattens. Hedrick leans on this paper heavily, so it's worth checking exactly what it does and doesn't say.
The paper's espresso-preparation method “followed the approach devised by Lance Hedrick,” and Gagné helped shape it — so Hedrick genuinely contributed to the work he's pointing to. That's worth knowing, and to his credit; it doesn't change the physics.
Correct. Flow averaged over the last 10 s of each ~120 s run. They report basket pressure, calibrated down from pump pressure — which matters (below).
The data points peak near 4.5 bar and plateau — which is exactly how Hedrick describes it too.
The paper is descriptive, not prescriptive. It never says “brew at X bar.” Gagné's physics is that more pressure buys less than you'd think, and that above ~10 bar the pores collapse — yet he still publishes recipes that run near nine bar. So nobody here is claiming “six bar beats nine”; the physics just says the top of the range is doing less than the number implies. And Hedrick's point, once you get past the video's title, is narrower than a verdict: that nine bar “never had an actual, provable, scientific nor historical reason,” and that the science suggests it may not be ideal.
A DE1 can pressure profile, and it logs every shot, so our community has been testing this exact question for years. Three things from our community forum are worth putting next to the physics.
I lowered our default max pressure — for exactly this reason. When I released the Adaptive profile v3, I dropped its maximum from 9.5 to 8.6 bar:
“9.5 bar is the absolute max that a puck can take before it starts having a secondary compression, and flow going down as pressure increases. Generally this makes for worse tasting coffee ('muted')… Also: 8.6 bar is experimentally what traditional boiler machines running at 9 bar actually delivered to the puck — I discovered this reading the SCA competition machine setup process (9 bar actually reading at the puck as 8.6 bar).” John Buckman, Decent Espresso
That last point quietly corroborates the 2026 paper's basket-pressure correction: a machine set to “9 bar” isn't putting 9 bar on the coffee. The gap between the pump and the puck is real, and about the size the paper calibrates.
But it depends on your grind, and we don't all agree about it. Gagné — here as a DE1 owner, not just a physicist — hasn't seen the same wall I did:
“I have not noticed a 'secondary compression' at 9 bar with the SSP ultra low fines (the point where bed resistance shoots up).” Jonathan Gagné
That is precisely the variable Hedrick found dominated his own testing: with a low-fines grinder and a light roast, the whole picture shifts to lower pressures and the high-pressure wall softens. Two careful people get different answers because they use different coffee and grinders. That's not a contradiction — it's the honest shape of this question.
And “pressure” is only one of three levers. Collin Arneson lays out the framework the argument usually skips:
“Flow rate during the shot is determined by three things: the changing viscosity of the fluid inside the puck (resistance goes down as solubles leave); fines migration (resistance goes up); and the inherent resistance of the puck from its particle-size distribution. How the first two interact is what's behind the differences between grinders and coffees.” Collin Arneson, Decent Espresso
Set the pressure wherever you like; these three are what actually move the flow. It's the same conclusion Hedrick reaches from the other direction — that pressure between seven and ten bar can be described as “a red herring.”
At what pressure does this community actually brew? Because DE1 shots log to visualizer.coffee, I sampled 141 recent public shots and took each shot's sustained peak pressure.
The pattern is clear: most of us aren't brewing at nine bar, and almost nobody goes past 9.5 — which is right where I found the puck starts to compress a second time.
Put Hedrick, Gagné and our own shots side by side and the picture is clear. On the big questions they line up; the disagreements are narrow, and honest.
You don't need to chase nine bar. What matters is even flow after peak pressure.
Nine bar is a number we inherited from the stiffness of a spring, not one anyone tested their way to, and flow stops improving well below it. So if your shot only reaches six or seven bar, don't worry — you may actually be in a more forgiving spot. And if you can change your pressure, this is an easy thing to try for yourself: pull the same coffee at six bar and at nine, each one dialled in, and taste them side by side. It's the kind of comparison that led me to cap our own Adaptive profile at 8.6 bar, after months of testing.
Waszkiewicz, Myck, Białas, Puciata-Mroczyńska, Dzikowski, Szymczak & Lisicki, “Under pressure: poroelastic regulation of flow in espresso brewing,” Physics of Fluids 38, 063113 (2026). arXiv:2512.21528. Open data on GitHub + Zenodo.
An Espresso Profile that Adapts to your Grind Size (the √pressure law and the ~10-bar wall); A Study of Espresso Puck Resistance (the puck compresses like a spring); Extraction Uniformity and Channeling (why “fastest flow is always best” fails on a real puck). Books: The Physics of Filter Coffee, The Physics of Espresso.
Corrochano et al. (tortuosity/consolidation, J. Food Eng. 2015, which also quotes Petracco & Suggi Liverani 1993 at ~5–7 bar); Mo, Navarini et al. (swelling, 2022); Mo, Navarini, Ellero, Johnston (erosion, 2021); Matias et al. (swelling + erosion, 2020); Foster et al. (micro-CT infiltration, 2025).
Illy & Viani, Espresso Coffee: The Science of Quality (2005); Climpson & Sons, 9-bar-vs-6-bar blind test; Scott Rao / Decent Espresso on turbo shots and declining-pressure profiles.
John Buckman on Adaptive v3 max pressure; Jonathan Gagné on grind-dependent secondary compression; Collin Arneson on the three drivers of flow. See also Why is the flow curve shaped differently with different coffees or grinders?
The video this piece responds to:
john created 2026/08/25, john updated 2026/08/26