I was discussing espresso with a coffee enthusiast the other day and it reminded me to post some of the results of a semi-formal study I did about espresso brewing for some barista training. The idea came about in discussions with a trusted coffee colleague about the importance of a level tamp and why that is so important. Most baristas know (are told) an even tamp is important for a more even extraction of coffee tastiness across the entire brew bed. This is 100% true, as far as I can tell. What I didn’t realize is how uneven tamping can cause water temperatures to vary widely intra-brew, creating wide ranges of spent espresso puck temperatures. So I set out to record some data in a more formalized way to see if I could detect any patterns and come to some conclusions to share with the wider coffee community for confirmation/rebuttal.
Mis en place for my project included a portafilter, a freshly spent espresso puck (still in the filter basket), a camera and an infrared thermometer. Every shot followed the same protocol:
- 20 gram dose of the same coffee
- Two different grinders used (to ensure it wasn’t a burrs issue)
- 30-32 second shot
- 31 grams liquid
- 200F PID on both groupheads of a new (built Nov 2010) La Marzocco Linea MP2
- VST baskets + TxSpresso screens (ostensibly for a “straighter” wall of water. More on that below.)
- Spouted and bottomless portafilters tested
- Tamped unevenly on right side, left side, near side and far side for consistency

Even much less exaggerated off-tamps show fairly wide temp range discrepancies.
With those baselines established I set about seeing what, if anything, there was to see. I pulled a shot, took out the portafilter and took temperature measurements as quickly as possible. I also made sure to hold the laser as close to the spent puck as possible for a more accurate reading since IR laser thermometers lose accuracy with distance. The picture above shows a wildly exaggerated off-tamp, which is how I tamped for the experiment in order to get the full effect of the phenomenon. In real life the experienced barista would never tamp this way, but I felt it was the best way to produce a noticeable phenomenon of temperature variance. With more even tamps, the temperature variances are still noticeable–though not as widely varied–using the same protocols above (minus the uneven tamp, of course).
Some semi-random observations that I hope to draw out more in time, and come to a better understanding of…
…The puck temperature is hotter in the harder-tamped areas. I postulate that this is because there is more water built up in the headspace of the shot. (Incidentally, with a spouted portafilter, I seem to notice that in unevenly tamped pucks, coffee will often begin flowing from the harder-tamped side/spout first, then from the other spout. I don’t know why this is, since spouted shots seemingly get “averaged” by being forced to flow through the center hole before splitting into either spout. The uneven pour could be a result of a highly divided shot that literally clings to the side of the underside of the portafilter and down its respective side of the center hole and then into its respective spout side. The evidence of this comes when one observes uneven development of the shot in the bottom of the basket on a spoutless portafilter. Coffee begins beading up on the outer rim of the harder-tamped side first.)
Expel the puck and the temperature variances seem to be the same on the corresponding bottom side of the pucks.
As a semi-related thread, I postulate that with a TxSpresso screen water is debited in a straighter wall downwardly, through the puck and out the bottom of the VST basket (again, in a straighter wall downwardly). I believe this is due to the precision, centered holes in both the TxSpresso screen and the VST basket. A traditional dispersion screw/screen is more wishy-washy and better conforms to the slight arc of the grouphead block on La Marzocco machines, forming what appear to be teardrop-shaped waterfalls. TxSpresso screens/screws, at first blush, would seem less well suited to the slight arc of the LM grouphead block, building up water irregularly in the completely flat, rigid screen, versus conforming to the grouphead block’s slight arc (and thus debiting water relatively poorly to the puck). This seems true until one realizes that a) the TxSpresso dispersion screws have more holes (four versus two) to relieve any such pressure; b) the snugness of the TxSpresso screen, while lying flat across a slightly arced block, actually disperses water more evenly and possibly more efficiently (i.e., a straighter waterfall) because it fits slightly more snugly onto and around the grouphead block versus a traditional dispersion screen. This is evidenced by the fact that there is far less coffee build up when one removes a TxSpresso screen at the end of a shift versus the traditional screen. And finally, c) there may actually be more hole space in the tiny TxSpresso hole array versus the relatively gaping holes of the traditional screen, allowing more relief of any supposed built up water pressure from the flat-screen-on-arced-block phenomenon. Thus, water can be hypothesized to flow more directly down into the coffee bed. The same principle theoretically applies to the lasered VST basket.
Back to temperatures. Below is an example (an actual reading) of one of the pucks taken for the experiment. It is actually the reading from the shot that is shown in the picture above. There is some variance in the absolute high and low temperatures, but the variances tends to stay the same unless a pinhole sinkhole develops somewhere in the puck. In that case, the sinkhole typically registers the hottest temperatures.

Sinkholes tend to develop near the outer rim of the basket on the boundaries of the harder-tamped areas–not usually in the low spots themselves. I hypothesize this is because water, which has to follow the path of least resistance, finds the first less-hard-tamped area it can and aggressively pursues its downward path there, forming a sinkhole. There were sinkholes in approximately 25% of the 40 different pucks I logged data for.
So that’s it for now. This is not intended to be some new area of study. Rather, in my mind, at least, it tends to confirm the edict in the coffee canon that one must tamp as evenly as possible. Somehow, while it is fashionable to use the latest and greatest grinders and espresso machines, it is considered gauche and novice to consider a more uniform tamping approach such as would be provided by a Swift grinder or one of those hammer tampers. (See? You thought, ”How gauche!” Didn’t you? As though using those tamping implements were somehow less artful, versus, say the most technologically advanced espresso machine. No one proposes going back to non-temp-stable espresso machines, for example, do they?) But I digress. The facts seem to be there. And it is the point of this blog post simply to make you aware of the facts on the ground. If anyone has read any work in this or a related subject, please feel free to point me in the right direction.