I don't think "fines are more than fine"

Q&A
1

Barista Hustle recently released a video quickly showing the results of an experiment. (here) But I don’t think the conclusions they drew from the test were appropriate at all.
If you haven’t watched the video (you should its like 30sec) they sifted coffee into two groups: >400um and <400um. Then they poured the same temperature water over both sizes and took samples at 30 and 60 seconds. They measured the TDS and graphed it all pretty like.
Then, based on the data, they said “fines are more than fine”

Here are the issues I have with this test:

I am not sure what their sample size was, but the video suggests they only had one data point for each size and time. This is by no means conclusive, and, therefore, conclusions should not be made, period.
Secondly, when is <400um considered a fine? I am assuming this video was aimed at espresso (based on the time samples were taken). Here Barista Hustle reports the mean particle size for 3/4 grinders dialed in for espresso is below 400um. All this video suggests is that you shouldn’t be using filter grind for espresso, and collection of grinds less than 400um extracts quicker than a collection of grinds larger than 400um. Basically: finer grinds extract faster than coarser. Whoda thunk.

So there’s my two cents. What do you guys think? Am I wrong? What conclusions can be drawn? Obviously I have strong opinions, and I apologize if I came off harsh or arrogant.

2

You’re not wrong.

It was a “cutesy” test :smiley: Helps generate some social media conversation.

Still, there is this phobia about “fines” which I think the point of the experiment, was just to push back on that phobia a little.

For brewing, possibly ignore the fines and instead filter out the boulders first… one of my experiments in the future with the Kruve. I would like to do some experiments putting those boulders back through the grinder and sifting again… anyways…

2 Likes
3

I think @homecafe is right about it being a fun video to generate some conversation, as well as trying to reduce a “phobia” about fines. I don’t think it was meant to be a vigorously scientific examination. You are correct though, that the video alone seems to only prove that finer grinds extract more quickly in a fixed amount of time.

As for this video sparking conversion, I think that Matt Perger’s explanation of the grinder paper he co-published gives some good context. This is the more vigorous science that I think gives the conclusion of the video credibility. I think they had their conclusion already, and that the video illustrated it.

According to Pergers’s summary of the grinder paper, the smallest grinds in a given profile contribute most grind surface area – approximately 70%. More surface area means more contribution to flavour. And if 70% of a coffee’s flavour comes from the finest grinds in a profile, then every cup of coffee is mostly defined by the finest grinds. If fines were inherently bad, then probably every coffee would be bad. If many coffees are tasty, and the flavour of those coffees is primarily determined by fine grinds, fines must be fine. Nothing to be afraid of, necessarily.

I think this is the idea they are trying to promote. (Of course, I may be wrong.) As far as defining fines goes, I think their point was more that a finer spectrum of grinds (<400um, approximately the modal grind size of a dialled in K30 for espresso) could produce tasty brewing results, not so much giving a specific definition of fines. Not so much a warning against a “filter” grind for espresso, but perhaps a warning that you shouldn’t use a “filter” grind for filter, as coarse grinds under-extract compared to the most prominent grinds/flavours in the brew (e.g. the idea of removing boulders with a sieve).

That’s my take. Forgive me if I over-explained anything. Better to lay out the whole argument and explain something you already know than to assume we are on the same page while we are not. I think I said pretty much the same thing as @homecafe but with more length… haha.

Wes

1 Like
4

It is odd that many people voice concerns about how many “fines” their grinder produces, compared to the number of folk who are worried about how fine their grinder is set. If you want less particles under 400um, set the grinder coarser, if you want more set it finer. Fines are inevitable, every unsifted brew, good & bad, has them.

The problems I have with the video are:
Ask different people what fines are & you get different answers. We can’t really discuss the role & impact of a thing we cannot define.
Kruve, mesh sieves & LPA may all show a different relationship of sub 400um particles, so can’t simply replace each other, nor can we use “400um” as having the same meaning across all conventions.
People only talk about fines & boulders, this is a bit oversimplified for me & sounds like a line out of a movie, “There’s only 2 kinds of grinds that come out of a grinder: fines & boulders…which one are you?” :slight_smile: What is the rest of the distribution made up of? Is there no normal range/deviation around the mean?
In percolation the smallest particles help control flow & in turn, extraction. If you grind coarse (less particles under a given size at the smaller end) you can control the flow & normalise the extraction. If you sift out the largest 10-20% you can extract more without over-extracting…but hang on…we were removing the particles that extract least & effectively increasing the component that should be contributing over-extracted flavours? Those large particles that skewed the brew with over-extracted flavours seem to be the same ones that contribute the under-extracted flavours too.
In immersion, sure they extract faster, but then are hard to separate from the beverage, whatever the level of extraction…an unfiltered 30 or 60 second steep is going to be too hot to enjoy & full of sediment. Waiting for it to settle, or filtering it takes time…so we’re back to limiting the silt we put in in the first place, longer brews & coarser grind. If you grind fine enough to hit a 24% immersion extraction in reasonable time, then take out the smallest 10-15% by mass,what happens? I haven’t seen EY drop for French press brews by sifting out the smallest component, but they do taste much cleaner & sweeter. A method for successfully & quickly filtering very fine immersions could be a game changer though.
In both drip & immersion, very fine particulate can increase bitterness/decrease sweetness, just by their presence in the cup. Are we confusing this with a bitterness we think is solely due to dissolved content & perhaps missing the obvious?
My perception of fines is that they are outlying, small particles,(boulders outlying large) if these only make up 1-5% of the mass of the dose, they cannot contribute 70% of the flavour.

4 Likes
5

Let me clarify: this post is not about the virtues/vices of fines. It’s about the experiment. I think the conclusions BH drew from the experiment are completely not applicable to the scope of the test. BH is basically spreading misinformation. What they said could be true, but nothing they did in their “experiment” can prove or even suggest that. They are spreading an idea, true or false, under the guise that is true and proven by science.

(I do like what Mark said about fines not conributing 70% of the flavor even if they have 70% of the SA. The video, ironically, does suggest this. Just look at the extraction numbers; <400um is not >70% of the extraction. But again, not related to the original post)

1 Like
6

Hey @MWJB! I would agree that it is important to focus on grind size for desirable results. But a grinder that produces more fines and produces grinds closer in size to fines will probably give better results (e.g. EK43). Why not look for that style of grinder?

And while “fines” does mean different things to different people, Maxwell Colonna-Dashwood (who worked on the grinder paper with Perger) has defined them as being between 0-80um in size. I do think we can discuss their role and impact, at least within Perger and C-D’s framework (which seems like a reasonable one to me).

You raise some good points in terms of equipment and conventions used, but once again I don’t think that the “Fines are Fine” video was meant to be a thorough, scientific analysis. In the grinder paper they get into laser diffraction particle size analysis, which is probably more reliable than the stuff we can typically use in a shop.

While we do focus a lot on fines vs. boulders, the fines are much more important to flavour than the boulders. Fines contribute so much to a cup’s flavour because of their massive surface area, which is approximately 70% of the total surface area. This is important to distinguish from the portion of the dose that they make up by weight. Grind-for-grind there are waaay more fines. Surface area contributes more to flavour than proportion by weight in the dose.

The issue is evenness. Your coffee’s flavour is “maxed out” at the extraction of the finest particle. Basically any grind that is larger than your smallest particles is contributing to unevenness because, relative to the finest grinds, it is under-extracting. How much potential surface area is hidden inside the non-fines? What if we could extract that flavour into the brew? Then evenness would be increased.

If a recipe was already over-extracting a coffee, then any increase in evenness will only highlight this over-extraction. But if the recipe is giving good results, then more evenness should amplify that and give even better results.

Before the grinder paper, I think we generally assumed that the primary flavour in a coffee came from the modal grind size in a grind distribution by volume. Now we are more concerned with the grind sizes that contribute the most surface area.

By removing (or crushing) non-fines (including boulders), you create a more even brew, highlighting the flavour that the fines were contributing all along. The trick: a) not over-extracting; and b) figuring out how to brew fines. (Sounds like the French press wasn’t giving good results.)

At the end of the day, fines are present no matter what, and they contribute the largest portion of a coffee’s flavour. So how do we get the rest of the grinds to contribute to the same degree?

Anyway, that’s how I understand it. I don’t mean to argue for the sake of it. I think there are significant implications for deliciousness in all of this. I think that understanding fines better can help us improve how we brew coffee through increased evenness.

Wes

1 Like
7

Wes,

First, here’s this. Its Scott Rao’s view on fines. I think his definition of fines is enlightening.

Surface area does not inherently mean more extraction, but rather more rapid extraction.

You say any ground larger than the finest contributes to unevenness, but I would think that any grounds that extract much slower or much quicker (fines and boulders) than the average ground contribute to unevenness. This is basically the second and third moments in probability theory and statistics (standard deviation and skew). So using a grinder, like an EK, that pulls the average towards the finer particles, making the standard deviation smaller and the skew less severe, produces a more even extraction. Measured extraction is, after all, the average extraction of every particle.

So you could do as you say and grind to produce only fines. That would make the standard deviation quite small. However, I think sieving the fines (and boulders) or making a grinder that produces less is just as viable. Also, fines extract nearly instantly, so brewing with only fines without over extracting is a matter of tenths of a second.

8

Completely agree @MWJB. Even for espresso based coffee, I find shots with less solid particles taste “cleaner” and that it is easier to distinguish different flavours. People seem to get very hung up on under and over extraction, perhaps because they don’t really understand the dissolution.

Great to see some well founded criticism in this thread!

As for claiming the “< 400um” sample is “fully extracted” despite only having made only two measurements…

2 Likes
9

Yeah, I thought 30%-35% of coffee mass is soluble, so completely extracted should be 35%…?

10

@WesG all good points, I welcome the insights & discourse, but I’m not sure how 80um applies as a cut off for fines in that scenario (there are 3 ways of presenting LPA data, for some reason Colonna-Dashwood et al only decided to show 2 of them), nor what the relevance is generally? I’m aware that individuals have stated what they think fines are, surely there must be a universal definition, encompassing non-EK/non-espresso & relating to grind distributions in general? If your grind size is 800um average, then sub 80um is almost nothing as % of mass.

I think we have a habit of making assumptions then discussing these assumptions like they are fact, like fines themselves instantly over-extract, or are solely responsible for over-extraction. I think there is an element of ‘cart before the horse’ in terms of studies showing how much soluble material is drawn from what component of the grind distribution. I might like to assume that fines break up the way they do because these are areas with less, malleable, soluble solid content, thus prone to shattering into smaller sizes. I don’t know, just an assumption :slight_smile: . I also wonder, given that the cellulose in roasted ground coffee is a porous matrix, what can be observed in terms of changes in particle size pre & post brew? More than a quick experiment like the OP sure, but it would be good to correlate something that ties up particle size (within a distribution) & extraction.

Thanks, but my French presses typically give very acceptable results, the biggest potential pitfall I have identified is an excess of particulate in the cup. My point really being that removing the sub 400um doesn’t rob the result of positive flavours, nor significantly reduce EY (if 70% of the flavour was contained in this portion of the grind, surely it would be very apparent?).

@ntd.dillon The CBI stated more like 30% +/-5% (IIRC lowest around 24% via AOAC OMA & performed at MIT, not immersion EY as shown in video). However, 24%-25% is very feasible in a declining temp, immersion extraction yield (as shown in the video) & you’re not likely to exceed this greatly without maintaining constant slurry temp, or introducing a flow aspect (e.g. Clever, siphon). So I would agree that 24-25% immersion EY would be a full extraction in real world terms in this case, but agreed, not the absolute max possible.

1 Like
11

If we wanna talk about fines extracting rapidly :slightly_smiling_face::

Rao says:“There are two phases of coffee extraction: surface erosion and inner-particle diffusion. When water contacts a large ground-coffee particle, it rapidly dissolves coffee solids from the exposed surfaces of the particle. In perhaps a few seconds, such erosion is nearly complete.
Extraction from the intact cells within a ground occurs by diffusion, a process that takes minutes, and is never complete, at least under reasonable brewing circumstances.”
Using his definition of a fine (a ground with no intact “cells”) extraction should take seconds because the entire fine is readily accessible. Diffusion doesn’t have to occur.

12

Hey all!

@ntd.dillon — Rao raises some very valid points, however his definition of fines being fragments of cell wall isn’t all that easy to quantify. Anyway, regardless of what we call them, particles under 70um can apparently create up to 70% of all surface area on an EK43.

@MWJB is correct — I have made a leap by claiming that 70% of surface area correlates to 70% of the flavour. Thank you for pointing this out. There would certainly be plenty of flavour coming from the inside of larger grinds not accounted for by the 30% surface area contributed by larger grinds.

It would be interesting to quantify this somehow to discover how much more potential surface area is hidden within 80+um particles. If @MWJB is correct that <80um particles make up, say, 5% of the mass of a dose, then that could mean 5% of the dose (by weight) makes up 70% of the surface area. Does that mean we are hiding 95% of the dose (by weight) under 30% of the surface area? That would be pretty inefficient.

Water should easily dissolve compounds from the immediate surface of any particle, but then it has to work its way into and out of larger particles in addition to the time it takes to dissolve compounds inside them. Thus, the water will extract (presumably) more of the same compounds from the inside of a particle as the outside, just later on in a brew. It would be the surface area of all grinds that contributes to the primary & furthest extracted flavours in the brew, as the surfaces are immediately exposed to water. Anything tucked away inside grinds would be under-extracted relative to the outsides of particles of all sizes.

While I agree, @ntd.dillon, that sieving fines and boulders (whatever one’s definition of those are) is probably equally viable for minimizing standard deviation, I think there will always be unevenness in flavour between the inside and outside of large grinds. Imagine over-extracting the outside of a boulder, under-extracting the centre of a boulder, and then getting some great, balanced flavours from somewhere in between. Then amplify that unevenness until you have a full dose of those boulders. Unevenness seems to be an inherent property of large particles.

To eliminate the flavours that fines give off, wouldn’t you have to eliminate surface area altogether? Maybe reducing the number of fines helps to decrease the intensity of these flavours, but as long as there is surface area, there will be some amount of relative over-extraction if the no-fines-no-boulders approach is taken.

It would be interesting to test the idea that fines extract instantly. Surface area does mean more rapid extraction, but isn’t this due to availability and not solubility? (I.e. more rapid relative to the same compound hidden inside a grind). Don’t different compounds still take different amounts of time to dissolve? Won’t a complex acid still take longer to extract than a simple one, even from a fine? And how long do the compounds that give us “over-extracted” flavours take to extract? If it is tenths of a second, wouldn’t this be the upper-limit of brew time, whether brewing fines or boulders — as you will still extract these compounds from the surface of a boulder if not a fine?

I would want to see some more evidence before accepting Rao’s proposition that in “perhaps a few seconds” erosion of compounds from fines is complete. I could see this being true for undissolved solids, but isn’t there more to dissolution than moving particles from one place to another?

Great conversation. Can’t wait to hear back.

Wes

13

So… mulling this over a bit as I continue with this issue of “fines” and “boulders” and the effect on brew.

I have been operating on an assumption, that boulders have the same extraction “curve” as fines just protracted in time. In other words, boulders take longer to extract. So, if not given enough time, the boulder is under-extracted.

Yes, I am playing the role of Captain Obvious :smiley:

However, my “epiphany” is this and illustrated through boundary conditions. Let’s use a really coarse grind. I mean really coarse.

As in a whole bean is our grind size. So adding water for extraction, means that given enough time, the whole bean will be fully extracted and not over extracted…

Except, this is not true. What will happen is that the outer layer of the bean will be massively over extracted and the “inner” part of the grind will be under extracted if at all.

Flip size, is that a Turkish grind will not experience this difference in extraction between inner and outer, so somewhere in the middle, is the optimal size for uniform extraction through the grind balanced with practical realities of brewing.

So at what point, for optimal brewing is the grind size critical for allowing proper extraction of the “inner” and “outer” grind particle?

Achieving uniformity of the outer and inner part of the grind, this has a dependency on cellulose structure and density of bean of course, but still, uniformity of grind particle size, while important, also needs to be at a size small enough for uniform extraction at that individual grind particle.

This leads to a bias of grinding finer for brewing, though counter to my experience, since I typically have better brews with coarser grinds at longer extractions. The finer the grind, the more challenging managing water flow and distribution particularly with larger doses (ideal for me is 28g).

Of course, I could just sieve and test this, and likely covered in a tech paper already…

14

I wish I had the article on the experiment of the “fines” that I couldn’t book mark. Now I cant find it. Anyhow, the articles was highly scientific and basically stated that the fines will agglomerate/bind to form a larger particle. This larger particle of fines then extracts more evenly. The trick was not to disturb the fines such as in a Bodum. It worked for me.

15

I just re-read @WesG post, and realized I just repeated “Imagine over-extracting the outside of a boulder, under-extracting the centre of a boulder, and then getting some great, balanced flavours from somewhere in between.”

Size of grind particle needs to minimize the effect that Wesley describes but yet allow uniform flow of water through grinds.This becomes very problematic at finer grinds from a practical concern, channeling, and the water contact time issue.

And very much I believe this “Don’t different compounds still take different amounts of time to dissolve?”

I don’t think we account for water contact time appropriately. I am also somewhat unconvinced as to concentration of brew liquid restricting extraction ability which Scott references as a dominant influence. Hence, the one pour method for V60 which seems to be unduly influenced by efficiency and repeatability concerns.

Scott?

16

Hey @homecafe,

I think your bean analogy is a great illustration of what I was trying to say. You are certainly right, if we are correct about finer grinds being inherently more even, then brewing is a matter of balancing practical aspects of an even extraction with fineness of grind.

As far as practical stuff goes:
With smaller doses I have been able to grind pretty fine for a Kalita 185, but it looks like I brew much smaller cups than you do (14g coffee to 210g water), so my dose would likely create much less resistance than yours, allowing me to grind, relatively speaking, quite a bit finer. There are a couple other things I have found can help, though, the major one being pouring all of the water immediately (as opposed to multiple stages - another advantage of brewing a small amount of coffee) and the other being hot water temperatures (100C/212F) to increase flow rate (but this also influences rate of dissolution).

As far as channeling goes, I find that giving the grinds a stir and then tapping the brewer downward one time will usually give a good, flat bed with no channels. My brews usually come out in a reasonable 3-4.5 minutes, depending on the coffee, water hardness, and temperature (I’ve actually been brewing at 204F recently to balance out harder water). Essentially I’ve just had to get creative with ways to change flow rate without changing grind size. It quickly becomes a game of balancing rate of extraction with flow rate.

Back to theory:
I think the important questions to consider are: how long does it take to extract the first “over-extracted” compound, and what is the discrepancy in time between when we start extracting the outer layers (& fines) and the inner-most portion of any grind? The answer to the first question will tell us how long our target maximum brew time should be (based on our rate of extraction, determined by things like temperature, agitation, brew ratio, etc.). The second will tell us about evenness (how well are we extracting to the same “maximum” from every particle in the same time?) and is quantified in grind size distribution analyses.

Unfortunately, I don’t think the video gives a clear answer, but I think there is enough evidence from the grinder paper to suggest that fines are fine.

I’m pretty convinced that “fines are more than fine” based on the paper alone, but also based on theorizing about how particle size correlates to evenness, and personally (although anecdotally) I have had results I am very happy with by applying these principles in my own brewing.

Have you made any progress applying these ideas to your brewing? Thoughts?

17

Hi @WesG

I have a few different drip cones, some require a more specific pour regime than others, like the Kalita Caffe Uno which is a tiny travel cone, forces small pulses & an accordingly coarse grind (7-12% at 400 with Kruve). At the other end of the scale the Bonavita & Brewista steep brewers which I can use as drip brewers with a finer grind (~25% @ 400 with Kruve), long bloom with valve shut, then open valve &all water in one fast pour (10sec). My own subjective assessment is that I cannot determine a preference in taste between these methods & anything in between. Finer grinds can produce a more syrupy mouthfeel, coarser grinds a thinner mouthfeel.

Brew time for these approaches can vary wildly, so I don’t see brew time as a good indicator unless brewer & recipe are specified. Standard deviation in brew time can vary from +/-10sec for one method to +/-60sec for another (single pour brews will have the larger StDev because the grind & larger weight of water above the bed will play a larger part in variations in draw down time, compared to short pulses with a coarse grind, e.g. if you pour in 20sec pulses and the bed drains out in that 20sec, your brew cannot be a minute longer than another). All these brews & methods have fines, just the coarse ones have bigger fines…all burr grinders produce fines.

For immersions, these always taste better to me with sub 400/500 removed, extraction yield doesn’t change greatly (unless the brewer drains through the bed, like a Clever & there you might drop 2%), so it strikes me the off flavours are more related to the small particles’ physical presence in the beverage, rather than extracted solubles. I’d be curious to see how someone would brew quickly with extremely fine grinds in immersion & still produce a clean tasting drink. Brew time targets will be less precise here, it’s unlikely in declining temperature steeps, without flow through the bed, that you will be hitting over-extraction at all.

Evenness will show in the average EY you get - if you’re finding drip brews average tasty EYs at say 22% (with a range of origins) instead of 20%, then your grind is likely more even. Commercial coffee producers do this every day with roller mills.

The grinder paper discussed espresso grind, the sub 70um percentage of manual drip grind (600-900um?) must be pretty tiny? But, agreed, I also think fines are fine for percolation (as long as they don’t get into the beverage, as they can more easily in large brews). For immersion…I’m yet to be convinced, that even if they can extract quickly, that this will correlate to as sweet & clean a brew until someone develops method for quickly & effectively separating the beverage from the very fine grounds.

18

Hey!

My experience with increasing fineness is that strength will be noticeably higher and whatever the “furthest” extracted flavour was prior to increasing fineness will be emphasized. I would agree that you have to specify a brew device and method to talk specifics, but I think the general theory still holds true. One would just have to apply it to specific situations.

In any context (water hardness, brew device, roast, temp, etc.) I think there would be a fixed amount of time in which it takes to extract the first “over-extracted” flavour, and by bringing the brew time to just shy of that we won’t over-extract. If you’re brewing with hotter water, or harder water, or more agitation, this time could be shorter, but that’s because the context has shifted. It sounds like we’re on the same page there.

Not sure about the French Press in particular. I have never cared for particles in my coffee, so I can see why more fines in a French press might not produce the nicest results, but as you said that seems more like a filtration issue than an extraction one.

I have been typically getting tasty results on my Kalita in the 21.5%-23% EY range depending on the coffee, as long as my brew time is reasonable (usually tasting ideal in the 3:30-4:20 range — depending on all the stuff I mentioned above). That’s just with a (slightly modded) Porlex. I would love to run with this stuff on an EK.

I would love a super clean French Press. Have you ever tried using a different filter with an AeroPress, tightening the cap over a larger paper filter so that it covers the entire cap, including the holes in the side? I find this produces a very clean result for the AeroPress. I have unfolded and cut a Chemex filter into the four sections, wet it, and used it for that purpose. I find the cup comes out about as clean as a Chemex would. You can also get away with grinding pretty fine because of the pressure, and if you decide to push the time further than a conventional 1m-2m AeroPress time, you can pretty easily get interesting results at ~25% EY in around 4 mins. Not a French press, but still (mostly) immersion if you do the inverted method.

Also, just as a caveat for all of this practical stuff, fine grind sizes like to clump, so I make sure to stir very thoughtfully & intentionally during any pre-saturation phase to make sure all grinds are getting wet simultaneously. And the shorter the pre-saturation, the quicker the brew, so that can help speed things up if necessary.

19

Hi @WesG

Sorry, I should have been clearer, in all the drip scenarios I mentioned strength & EY were comparable, grinding finer/coarser normalised by how stretched out the pours are. For immersion (FP), grinding finer can speed up extraction, but at various grind sizes my brews all seem to average 22% across a range of origins…just the time taken to do this extends as you go coarser. I tend to grind just fine enough to get my desired extraction, going much finer to save a few minutes can be a double edged sword, if you then need to take greater care at decanting. To be honest, if I want a quick brew I go drip, I tend to use FP more when I have time to spare.

With Aeropress I have tried different filters - multiple paper, single paper, combinations of paper with Able fine, but not larger filter paper, I’m hoping the Prismo will help in this respect & give a better seal with stock papers. 25% immersion is my usual target for Clever & Aeropress (unsifted grounds). The cleanest French presses I have made were by grinding relatively fine (fine particles sink quicker than coarse) steeping long, let everything settle, discard oily/silty surface layer & little disturbance at decanting (Hoffmann no plunge method), no further filtration…still a tad hazy, but secondary filtration often seems to have adverse side effects, as there’s no bed to assist filtration, you still get some silt in the cup. If I’m settling for a more ‘slapdash’ cup, I’ll sift out the sub 400/500 & use an Espro.

20

BTW, a very clean cup of french press from “The Hoff” :smiley: 10 minutes! Fines settle naturally over this time.

Alternatively, Aeropress is immersion so I have achieved a “cleaner” cup by using a metal AND paper filter both. Metal filter first then paper filter, added convenience is that when pre-wetting the paper filter on top, it holds the metal filter in place so that it does not fall out when placing on aeropress (inverted method). Digression.

But also, a very interesting and great method for V60 is from the World Brewers Cup winner 2016 which turns on it’s head our conventional approach (40/60) method. He allows complete drawdown in between pours! This actually works.