So for the last few weeks, I’ve been addressing different issues regarding Brad Schoenfeld’s recent paper suggesting that an incredibly high training volume, far more than have ever been suggested or used by any sane human, give the most growth. I won’t re-examine the issues I have with it but you can read my first diss track and my second diss track if you’re not got caught up.
Rather, as discussed two weeks ago, I want to now look at the other papers examining the issue of training volume and muscle growth. As it turns out there are currently 7 of relevance, including Brad’s, of which 2 came out within roughly a week of his. Since I have a lot to cover, this will take 3 articles to address everything I want to say. First, a bit of a tangent and this will be a long piece.
Building Scientific Models
While imperfect, mainly due to the fact that scientists are only human, the scientific method is currently the best approach to answer questions about our universe. This is because rather than being predicated on faith, intuition or being told by someone in charge, it is predicated on testing a hypothesis. A scientists asks a question, designs an experiment and sees if the data does or does not support it. If it does that means little until verification occurs. One study simply doesn’t mean squat in isolation. Maybe it was a crap study, sometimes researchers fake data (see Andrew Wakefield and vaccines/autism).
Ideally for science to occur you get verification, with more than one study showing the same finding. If verification comes independently, from a different lab, that’s even better. It’s always a little bit curious when one lab seems to always find the same results time after time but another lab always seems to find the opposite result time after time.
And how those results almost always reflect the bias of the person or people running the lab. But when two different labs get similar or the same results studying the same thing, that’s good. Presumably they don’t share the same biases and this means that the finding is more solid. It’s even betterer if that lab thinks you’re an idiot and wants to prove you wrong but can’t. Boom, now the finding is that much stronger. Because if that lab actually wants to prove you wrong and can’t, well…
One of my favorite stories from ages ago. The issue of muscle fiber hyperplasia had been long debated for years. One lab said it happened, another said it didn’t. In a fit of intellectual honesty, the second lab sent one of their people to the first lab to do some of the data gathering. Basically they sent someone who thought the first lab was full of shit to help them do the science. And as it turned out the first lab was right. That’s how you do good science and I wonder when this concept got lost.
But sometimes a similar study finds divergent results. Now, this might be methodological, perhaps the scientists don’t know how to properly blind, describe randomization or register studies for example. Sometimes it’s an issue of population or specifics. A training study in young men probably should get different results than in older men. Or older women. Or women on birth control or with PCOS. Which is why you have to study all of them rather than assuming they are the same and why science often only inches forwards. So over time you accumulate all of these freaking studies, some of which agree, some of which show different findings, etc.
Now you build a model. By that I mean you now have to take all of the available data and come up with a single overarching model that can explain all of it. Or at least the studies that are methodologically sound (the problem being here that researchers tend to define studies they don’t agree with as being methodologically unsound. Well, unless it’s their study, then it’s ok for it to be a pile of shit.) But presuming that the data is all good, the model has to include it. And sometimes when you build the model, the divergent findings make sense.
A non-training example. For about 40 years there has been debate over whether metabolic adaptation occurs during dieting. By this I mean the extra adaptive bit that causes a drop in metabolic rate outside of the weight loss related bit. Understanding that most of the work was in overweight individuals, about half said it happened and half didn’t. I have endless review papers in this folder on my computer and it’s fun reading the ones who say it does and doesn’t happen, especially when they carefully pick different studies in support of their belief.
But when you put the studies up against one another, collected the ones that said yes and the one that said no and see what their commonalities are, invariably what happened was the ones that said no were in extremely obese individuals, for whom adaptive thermogenesis is often trivial. It’s only when folks get past a certain level of leanness that it kicks in to enough of a degree to matter. I even remember one study that indirectly demonstrated this, observing no major adaptation in the first half of the study (when the subjects were fatter) and an increase once they got beyond a certain level of leanness. Mechanistically this can be explained by leptin transport saturating above a certain blood level, thought to be 20-25 ng/dl or so.
So if someone is super obese and has a leptin of 40, the body doesn’t really sense it until they get leaner and leptin drops below that level. If someone starts with 20, they get hit immediately. Aha, the model fits the data and can explain all of it and there’s even a plausible biological reason for the difference. Science crawls forward another inch.
But that’s how you do it in science, you take the study results on a given topic, group them roughly based on who they are studying (i.e. comparing studies in complete beginners and advanced trainees is kind of pointless) and see what falls out of the aggregate data. One way to do this is with meta-analysis. Take all the studies, statistically massage the numbers and come up with a way to compare them using effect sizes or something. Or you can do what I’m about to do and just look at each one in detail and then compare the results at the end.
Building a Model of Training Volume and Hypertrophy
Because what I’m going to do over the course of 3 articles is examine the aggregate data on training volume and muscle growth. But first some qualifiers. I won’t be looking at the tons of low volume studies out there. There’s really no debate that there is a dose-response relationship with more volume and better gains up to about 10 sets per week with 10+ having been established as providing the optimal response. Some of the studies I will examine have groups that are below the 10 set cutoff but they usually ends up just supporting what is already known so meh.
It is beyond this 10+ sets per week point that the questions arise as, up until fairly recently, there just wasn’t enough data to draw any useful conclusions. But as of the writing of this in October of 2018, there are 7 studies (3 of which came out in the last few weeks amazingly enough) and this provides enough data to draw at least tentative conclusions. Yes, future data may add to this and, as above, the model should always be updated with that new data.
At the same time, if you find that the majority of studies are in overall agreement, it tends to be rare that new research will turn the model on its head or change it completely. At best it will refine it somewhat (perhaps adding data for a different group, so maybe older trainees need more or less volume or whatever) or let you draw betterer conclusions. I’m not saying it can’t happen, simply that once you have enough studies saying the same general thing, you don’t expect to see hugely divergent finding show up unless it’s in a completely different population or there is some huge change to the methodology being used. So if all of a sudden a new and more accurate way of measuring muscle growth showed up it might very well turn the model on its head since old results would stop becoming relevant or might be proven totally false (i.e. what if Ultrasound weren’t even measuring actual muscle growth…).
This has actually happened on the issue of muscle fiber conversion. From about the 70’s for like 30 years, research had said it doesn’t meaningfully happen but new methods and technologies of measurement have turned that idea on its head. With a better ability to identify muscle fibers types (specifically, Myosin Heavy Chain or MHC subtypes), hybrid fibers and transitions between them, the old conclusions basically got thrown out and it’s now understood that much more fiber conversion than previously thought can occur. So it can happen where a well-established model gets demolished by a change in technology or measurement techniques And when it does, the model has to be changed. And until that occurs, all we can do is look at the current data with the current methodologies to build the best current model that we can.
I’m also not interested in beginners. There are a zillion studies on this already and, often, you see no meaningful difference in the responses to training almost irrespective of volume or anything else. One set or three sets, 60% of max or 90% of max, twice a week or three times per week, it’s all about the same. A lot of the gains early on are neural, learning to lift the weight basically, and it all just cancels out. Mostly what these are looking at is whether tripling training time or increasing training days from 2 to 3 is worth it for the average individual. And it’s usually not as any slight gains are far out of proportion to the increased training time commitment.
Yes, fine, sometimes there are small differences in results, more volume gives more practice and improves neural effects or whatever and you see a little bit more strength gains for 3 vs. 1 set or three days per week versus two or whatever. But it mostly equals out and isn’t relevant to the question at hand. The volumes are low to begin with, beginners all respond more or less the same and that’s not the question I want to address here.
Rather, the question is whether or not intermediate or advanced trainees needs higher volumes for optimal growth. Even the original Wernbom review, that I’ve often used myself but which a lot of people like to dismiss of late, had limited data on intermediate or advanced subjects and it was unclear if those numbers (~40-70 reps twice a week per muscle group) applied to intermediate or advanced trainees. But that was then and this is now and we have at least a decent research base to examine that might shed more light on the topic.
So the focus of the papers I will examine, 7 of them, will be on studies that examined volumes above 10+ sets (often compared to lower volumes) and that were done in individuals who were not rank beginners. There is one exception, a study I will include that used military recruits new to strength training but you’ll see that it doesn’t really matter in the big picture. Some of them compare two different volumes, some three different volumes, one examined escalating volumes over a number of weeks. But they are all fundamentally examining the issue of training volume and hypertrophy response and can be used to build a model of this to see what falls out.
So, as it turns out, this week Eric Helms analysis of Brad’s paper and review of studies on training volume came out in his MASS newsletter as well. And in it 11 studies were cited. And I want to explain why I’m only including 7 of those 11 because I feel that complete transparency is always the best approach. This is unlike the gurus who wait until they are backed into a corner over something to come up with a new argument.
Two in his list ( Paulsen, Ronnestad) were on beginners and I don’t care. One (Baker) was in recreationally trained men but trained only pecs, delts and biceps for three exercises three times per week and either 1 or 3 sets of each exercise (so 9 vs. 27 sets/week for the three muscles trained although Eric reports it as 9-12 vs. 27-36 probably due to differences in set counting). It used skinfolds for body composition, only reported the change in skinfolds without body composition data per se, did no direct measure of muscle growth and, in any case, both groups got the same strength gains (bizarrely the 1 set group had a bigger drop in skinfolds than the 3 set). I can’t be bothered to address it in any more detail than that so I’m ignoring it. It would not materially alter anything in my analysis, especially given that it found that higher volumes provided no more gains than lower. The fourth (Gonzalez-Badillo) was a study in Junior Olympic Weightlifters and looked only at strength gains in the competition movements. Humorously, it found that moderate volumes were superior to high or low but I’m still not including it.
But those 4 studies do not fit the criteria I laid out above to begin with regarding volume (10+ sets), training status (trained) or end point (hypertrophy). I just don’t want anybody who has read both thinking that I deliberately left out relevant studies. But that’s why I was specific in my criteria above ahead of time. If anything, those studies would all support my overall conclusions but since they don’t meet MY criteria, I’m not going to include them after the fact because that is intellectually dishonest.
Ironically, in the issue itself there is a piece by Greg Knuckols on how to trust studies and talks about how study pre-registration is important, to avoid P-hacking and throwing different statistical methods at the paper until something sticks, how not blinding a study with subjective methods is a problem and a bunch of others. You know, the exact same criticisms I brought up with Brad’s study that went unaddressed or were simply excused because “I don’t know science” Well, James, I guess Greg must not either because his article indirectly says that your study is methodological shit, too. It contained 17 total points and depending on how you count Brad’s paper failed 7-9 of them.
Despite that, Eric then concludes that Brad’s study, despite literally breaking about half of the points that that Greg said is important, was methodologically sound and based this on “I do research”. Basically just an appeal to authority and the “If you don’t do science you don’t get an opinion” like James used Eric also uses the “Since other studies have this flaw, it’s ok that Brad’s does” argument. It’s really sad to watch.
Eric also says that the authors of the paper addressed the online criticisms which is an outright lie. Brad never addressed a single question, Eric deflected with the 9 year old blog bullshit (he wasn’t an author) and James attempted feebly to defend it (at least he tried). NOBODY has even attempted to address Brad’s LIE in the discussion. NOT A SINGLE PERSON.
Eric also repeats the Bayesian factors, pointing out (honestly) that the Bf of 1-3 is weak (it’s NOT WORTH MENTIONING) And yet concluding that the study still showed a trend for the highest volume to be better. It’s odd, almost like he can’t make up his mind whether he’s evidence based (in which case the Bayesian values are meaningless and don’t support the conclusion) or just trying to stay on Brad’s good side by trying to defend the paper with weak sauce apologism.
Perhaps the most desperate argument was data that James ONLY put up on his BLOG that was NOT in the paper (so it doesn’t count since nobody else had access to it until it was necessary) that found an almost significantly greater volume load for leg extensions in the high volume group. OMG, this changes everything. No, wait, it doesn’t. I mean, does anyone give a flying fuck that the leg extension load volume (weight * sets * reps) was ALMOST higher in the highest volume group? This is a study primarily about fucking MUSCLE GROWTH and this is the best Eric or James can do? Fucking leg extension training volume? Jesus.
Oh oh wait, one more: the letter from the editors states that Eric provide an UNBIASED look at Brad’s study to address online chatter.
Unbiased my hairy ass. Nobody is unbiased, including me and true objectivity cannot exist. I’m simply the only one honest enough to admit it. I have my biases, I just try to put them up front so you know where I’m coming from. Everybody else ignores their biases and pretends they aren’t biased. But I guess Eric pulled out his Fair Witness cloak (get THAT reference, nerds) to write his piece. It’s the only explanation.
The same intro mentioned that the paper was published in MSSE, one of the top journals in the field. An appeal to authority and irrelevant. The Lancet, a top tier medical journal for over a century published Wakefield’s autism study so clearly even good journals can publish utter shit. This is just pitiful.
At least Eric made the rational conclusion regarding weekly sets (spoiler: NOWHERE CLOSE TO BRAD’S NUMBERS) but it was wrapped in nonsense and vagaries and apologism to sort of explain away why Brad’s numbers were so far out of the reality we live in without just saying it was a piece of shit paper with shitty methodology and shittier statistics that in no way supported Brad’s shitty claims or conclusions. But when seminar money is on the line I guess you pretend you’re unbiased and evidence based and hope nobody notices how full of it you are.
Unfortunately for him, I noticed. Stay tuned for my lesson on integrity for him.
I could literally make this another diss piece about Eric’s apologism for Brad’s study and just tear his MASS article to shreds but I already dropped the mic on that topic. So….
Back to the Point
I will be looking at the 7 relevant studies on the topic in some detail and will do so in their chronological order of publication. I will not really be comparing them as I go. Rather, that final analysis, looking at the body of literature in toto (No, not this Toto or that Toto) to see if any patterns arise will have to wait until Part 3 when I’ll also look at them in a slightly different way related to the next topic.
I’d note that there is still little to no data on truly advanced trainees at this point, perhaps 4+ years of consistent training. Or not enough to draw any real conclusions. We can quibble about what defines a true intermediate or advanced. Most of the studies had training ranges of 1-4 years or a minimum of 1 year of weight training with the one exception. I’d say at 3-4 years someone is advanced intermediate approaching advanced. Most of the strength levels were distinctly advanced beginner or approaching intermediate. So this still isn’t in highly trained individuals. But it’s the data we have.
And the specific question I will be examining is what the relationship between weekly training volumes (set count) and muscular hypertrophy are. As noted above, the original meta-analysis only concluded that 10+ sets was optimal but the data didn’t allow for conclusions beyond that at the time. So we ask:
- Does more volume keep generating more growth indefinitely?
- Is there some upper limit where the growth response stops or even reverses?
- Can we try to define some optimal volume where growth is maximized?
- Does that optimal volume possibly differ for different muscle groups (or upper vs. lower)?
The only way to determine this is to look at ALL the studies in some detail to try to build a model which is why this will take a solid 3 articles.
A Note on Volume Nomenclature
Since it will come in Part 3, I want to talk about how volume is being counted or considered. In many studies, only compound exercises are used although triceps, biceps or quadriceps size (either by vastus lateralis or rectus femoris) are all that is being measured. Others use a mix compound and isolation. This raises the question of how to consider the set count. Does one set of cable rows or bench press count as one set of direct work for biceps or triceps respectively? Should it be counted the same or differently as a set of biceps curls or triceps pushdowns?
As I look initially at the data, I will count it as if it does, primarily because this is how Brad Schoenfeld and his group have done it in their studies and meta-analyses and it makes the most sense to use a consistent (and their) method throughout at least initially. So any exercise done involving any muscle is counted as 1 set for that muscle and any other muscle that it works. A set of bench press is thus counted as one set for chest, shoulders and triceps. A set of rows or pulldowns is one set for back and biceps and a set of squats or leg press is one set for quads. A set of triceps extensions is only triceps, a set of curls is only biceps, and leg extensions are only for quads. I’ll say right now that I do not agree with this approach but I will go along with this nomenclature to maintain consistency with their original analyses and their own recent study until I re-examine it at the end a little bit differently.
And with that out of the way, let’s start with the paper that kind of kicked off this whole mess to some degree. I have linked the title of each study I will be examining to the Pubmed reference so that if anybody wants to check and see if I’m honestly reporting the data or not, they can. I am quite sure I have flubbed a number or two in this series since I write and type quickly but you can spare me the claim that it was deliberate. Usually it’s an obvious typo and I’m too lazy to check it that hard.
The first paper I want to look at is titled The Effect of Weight Training Volume on Hormonal Output and Muscular Size and Function by Karl J Ostrowski et. al. which was published in the Journal of Strength and Conditioning Research in 1997. I feel like I’ve talked about this a zillion times but will examine it again here.
It recruited 35 men who had been weight training from 1-4 years (hence not untrained) with a minimum of 1 year of regular weight training experience. They needed to be able to squat and bench at least 130% and 100% of bodyweight (we might quibble how trained they were but they were not beginners in any case). More specifically, the average squat 1RM was 133 kg (294 lbs) at a bodyweight of 77kg or 1.7xBW. Bench average was 87.5 kg (192.5 lbs) for a 1.13 x BW. This puts them at a solid intermediate squat and somewhere between a beginner and intermediate bench based on the standards you typically find online.
Each followed the workout that appears to the left (click to enlarge) and performed 1, 2 or 4 sets per exercise. The study lasted 10 weeks and size changes in the triceps and rectus femoris (representing quad growth) were measured via Ultrasound.
8 subjects dropped out for reasons unrelated to the training, leaving only 27 subjects but each group still had 9 subjects so it was balanced (although underpowered in a statistical sense). There were no significant initial differences between groups in anthropometric variables, training history, performance, or hormonal variables prior to training (and they were listed individually for each group which is fairly standard, or should be). So the groups were also balanced in this regard.
So let’s add up the volumes. Legs are easy, there were three quad and three hamstring movements once a week for 1,2 or 4 sets apiece and this yields a lower body volume of 3, 6 or 12 sets per week. We might question if this is truly high volume but it is looking at increasing volumes in a dose specific way and that’s what they did so that’s the data we have.
For upper body it’s a bit more complicated due to the presence of a push, pull and arm day. Using the counting convention I described above (where both compound and isolation movements count as one set for all muscles involved), and focusing on triceps, we count the pressing day as having 4, 8 and 16 sets for triceps (4 exercises * 1, 2 or 4 sets). The additional 3 triceps movements on Day 4 add another 3, 6 and 12 sets. Totaling that up we reach a triceps volume of 7, 14 and 28 sets per week for the different groups. I hope that makes sense. The same would hold for biceps but it wasn’t measured so it doesn’t really matter.
First I’ve presented the results for the rectus femoris indicating the starting and ending size, change and percentage increase (calculated by dividing the starting value by the change and multiplying by 100). I’ve indicated changes based on total weekly sets. This value is in mm^2 (millimeters squared, explaining why they are so much higher than the triceps values below)
|Quads (pre)||930 mm^2||940 mm^2||860 mm^2|
|Quads (post)||993 mm^2||987 mm^2||973 mm^2|
As I’ve discussed previously, this data is a little bit weird with the moderate volume group growing less than the low volume group although it’s probably just a statistical blip of zero relevance and it’s probably that in this group low volume and moderate volume were about the same in terms of the growth response. Percentage wise, the higher volume group got about double the growth.
However the researchers found none of this was statistically significantly different between groups. Not the starting values and not the changes regardless of volume. Still it’s hard not to see a trend from the 3 and 6 set groups and the 12 set group where 3 and 6 are the same and 12 sets got more (about double) the growth. Given what we know about the dose response to training volume, that is up to 10+ sets it keeps going up, this makes sense. In somewhat trained lifters, you might expect a higher volume to be superior or required for optimal results. The lack of statistical significance is likely just a statistical blip related to the study being underpowered statistically. But let’s all agree that a trend is there, one that would pass a reality check with the real world of the gym.
Make no mistake, 12 sets is only high volume by comparison and it’s a shame they didn’t have an even higher volume group as that would have been informative. Perhaps 15 or 20 sets would have achieved even a higher percentage growth that reached statistical significance. But that’s mere speculation and without the data we can’t conclude anything. But for the legs there is something of a dose response relationship. Both low volumes grew the same, the higher volume grew better.
Moving to triceps, I’ve presented the same data. These are in mm, explaining the vast difference in absolute magnitude of values from the leg data above (mm vs. mm^2).
|Triceps (pre)||44 mm||43 mm||42 mm|
|Triceps (post)||45 mm||45 mm||44 mm|
|Change||1 mm||2 mm||2 mm|
As with quadriceps, the researchers concluded that these changes were statistically non-significant although there is an apparent trend towards between growth from the lowest to moderate volume. Given the set count this also passes the reality check and previous data. Growth improves up to 10+ sets and 14 is higher than 10 and 7 is lower than 10 so that fits.
Importantly, there was NO REAL FURTHER INCREASE ABOVE THAT. Basically, unless you consider a doubling of training volume and time to be worth an additional 0.1% growth, we can conclude that this study found a plateau in growth with the moderate weekly set count. Basically 14 sets outperformed 7 but 28 sets was not better. So here we have a first indication that there might be an upper limit above which no further growth is seen.
Over 10 weeks, Ostrowski et. al. found that trained men showed a trend (non-statistically significant) for better growth at the highest volume (12 sets) for quads and a plateau at a moderate volume (14 sets/week) for triceps with no further increase at 28 sets/week. Moving on. Of some interest, the highest quad set count and middle triceps set count are in similar ranges or 12 and 14 sets per week. Since this study didn’t examine even higher quad volumes, we don’t know if there would have been further growth above that point.
Dose-response of 1, 3, and 5 sets of resistance exercise on strength, local muscular endurance, and hypertrophy.
The next study is by Radaelli et. al. published in 2015 in the Journal of Strength and Conditioning research. It recruited 48 men from the Brazillian Navy School of Lieutenants. The subjects were untrained in the sense of having zero weight training experience but were experienced with “traditional military training involving body weight exercises, such as push-ups, pull-ups, and abdominal exercises”. This is reflected in their baseline strength values; the 5 set group had a 5RM bench of 89.6 kg at baseline which might translate to roughly a 105 kg max (assuming 5RM is 85%). The average bodyweight was 79.3 kg and this equates to a 1.3BW bench in this group. I’d note that the 1 set group was much weaker with only a 64.5kg 5RM, equating to roughly a 76 kg 1RM or less than bodyweight. The 3 set group was in the middle with a 73.4 kg 5RM and the control group was in that realm with a 68.3 kg bench.
The study lasted 6 months which would have taken them through the rank beginning weight training stage. So while this is still a study on “beginners” the results should be illustrative. You’ll see at the end that it doesn’t matter much.
The men were divided into four groups. The first did bodyweight calisthenics 3X/week for an hour while the others performed a weight training workout consisting of bench press, leg press, lat pulldown, leg extension, shoulder press, leg curl, biceps curl, crunch and triceps extension three times per week. Groups did either 1, 3 or 5 sets per exercise in the 8-12 RM range. Growth was measured only in the arm flexors (biceps + brachalis) and triceps. But here the volumes differed.
Again using the counting convention described above there are three exercises that work the triceps: bench press, shoulder press, triceps extension. So that’s 3, 9 and 15 sets done three times weekly for 9, 27 and 45 sets week for triceps. For biceps there are only pulldown and biceps curl so that’s 2, 6 and 10 sets per workout for 6, 18 and 30 sets per week. Quads would have been the same as pulldowns with only 2 exercises for 2, 18 and 30 sets per week.
First let me look at the overall body composition results since I think they are vaguely indicative of just what this study did not show.
I’m gonna focus on Fat Free Mass (FFM) and let me note that this is not a particularly good proxy for actual muscle growth since it can be impacted by too many things (i.e. water, glycogen, poo). It’s still worth examining. Because what we see is the following.
|Control||61.95 kg||64.86 kg||+3kg|
|1-set||67.24 kg||67.70 kg||+0.5kg|
|3-set||63.01 kg||65.99 kg||+3kg|
|5-set||71.4 kg||74.7 kg||+3.3kg|
So the calisthenic group gained 3kg, identical to the 3-set group while the 1-set weight training group gained almost nothing. And the 5 set group only gained a little bit more than either the control/calisthenics group although everybody stomped the 1-set group.
Ok, does this pass the reality check to anybody? Does anybody believe that an hour of calisthenics three times per week beats out even low volume resistance training in beginners training for FFM gains? Yeah, me neither.
Irrespective of that, it certainly appears that going from low to moderate volumes improves results (0.5 kg to 3kg) but going from moderate to high doesn’t do much more (3.3 kg). Basically, there is a seeming cap for volume and FFM gains, inasmuch as FFM gains per se are a good proxy for muscle growth. Which they really aren’t. But it already sort of illustrates how goofy the results are at the outset. And it gets worse.
Ok, moving onto the actual growth data. Unfortunately, the researchers chose to present the results in this tiny ass little graphic that is really hard to see and the numerical data was not provided so I cannot present it as I did in the study above. I could try to estimate the numbers from the graphs but then we have the issue of MY bias and it hurts my eyes anyhow. So I’m just focusing on changes and whether or not they were statistically significant or not. This is shown in the graphic by the symbols and shit over the bars. Click it to make it bigger and read the legend to figure out what was different from what.
Ok, so biceps is on the left and recall that that group was doing 6, 18 and 30 sets per week. Here were the results:
|Group||Control||6 sets||18 sets||30 sets|
|Change||None||None||Some (more than 6 or control)||Most (more than 18 sets, 6 sets and control)|
I guess this is kind of a dose response relationship although not really since it’s actually none, none, some, more. For a true dose response relationship, you’d expect zero for control, some for 1-set, more for the 3 set and even more the 5 set group or something.Ignoring that let me ask the following question:
Does anybody believe that, in beginners (well, newbies to weight training) literally ZERO biceps size was gained with 6 per sets week over 6 months of training? Because this actually runs counter to essentially every study ever on on beginning trainees where those volumes reliably generate at least growth. And yet this study found zero (or nothing that was statistically significant). Yeah, I don’t believe it either.
Now let’s look at triceps. Now we see an even weirder pattern.
|Group||Control||9 sets||27 sets||45 sets|
|Change||Zero||Zero||Zero||Some (different than all other groups)|
First and foremost, this isn’t a dose response result which as above described increasing results with increasing volumes. Second and secondmost, does anybody believe this? Does anybody believe that, in total beginners, 9 or even 27 sets per week generated ZERO growth in the triceps and it took 45 sets/week to see anything happen? Can this be rationalized against at least some biceps growth in the 18 set per week group (do triceps need 2.5 times as much volume to get growth?) Because again this runs counter to every study ever done. And none, none, none, tons is not a dose-response. The results seem random and don’t make an iota of sense.
I’d note that at least a plausible explanation is that the prior military training the subjects had done altered their response to training. It’s still difficult to reconcile the different responses in the groups and for the different muscles. It’s also important to note that body fat percentage went down in all groups by 4% in the 1 set, 6% in the 3 set and 5.3% in the 5 set group (the control group saw no change in bodyfat) so the subjects were in a deficit. This could definitely have altered the results although it still does not explain the overall pattern of growth in individuals new to weight training. But there was also that pesky LBM gain. So something weird is going on. It also still fails the reality check completely. Nothing in this study’s results make an ounce of sense.
In this vein, I want to refer back to a rather cool older paper titled Time course for Arm and Chest Muscle Thickness Changes Following Bench Press training by Ogasawara et. al. and published in 2012 in the journal of Interventional Medicine and Applied Science.
In it, 7 healthy untrained men performed nothing but bench press for 3 sets three times per week and they did it for 24 weeks, the same 6 months as the Radaelli study. Biceps, triceps AND chest muscle thickness was measured (making me question why chest isn’t measured more often since it’s clearly technically possible and would give way more information than doing a ton of chest and shoulder work and only measuring triceps). I’ve presented the results below.
So this is super interesting. So no shock, essentially no growth in the biceps. Good control here, you wouldn’t expect bench to have much impact on biceps growth. Triceps did not show significant growth until week 5 (the asterisk) and then more or less plateaued about halfway through the study. In contrast, bench showed significant growth by week 3 and showed a more or less constant gradual increase to the end of the study as did 1RM (which you’d expect as muscle size increased gradually).
So why am I bringing this up? This study was identical in duration at 6 months to the Radaelli study and used a very moderate volume of 9 sets per week for chest with no other confounds in terms of exercises being done. And it generated measurable growth in both the pecs AND triceps with nothing but that volume in rank beginners. And yet somehow Raedelli couldn’t generate ANY triceps growth until 45 sets were done. None. Zero. Zip. Nada. Sure.
Like I said, Raedelli’s results run counter to every study ever done on the topic along with common sense. Combined with the FFM data where the calisthenic group outperformed the low volume training and matched both medium and high volume weight training groups, there’s simply no reason to even consider the results as meaningful or correct. The data is seemingly random, makes no sense, there is no logic to biceps getting at least some growth from 18 sets but triceps getting NONE with 27 and needing 45 to get anything.
So forget this study.
This paper is a pile of crap so far as I’m concerned. The results are random and fail every reality check while contradicting every previous beginner study to date. I’m throwing it out and you can agree with that decision or not but I think that the results are garbage and I will be ignoring it going forwards. It doesn’t materially impact on my conclusion as you’ll see in Part 3.
The next paper I want to examine was done by Amirthalingam et. al and published in the Journal of Strength and Conditioning Research in 2017. It recruited 19 healthy males who had an average resistance training experience between 3.5-4.8 years although the standard deviations were pretty big (the 10 set group was like 4.8 years +- 3.9 years so it was a spread of 1-4 years or so). So they were trained, at least to varying degrees.
However, looking at bench press, the 5 set group had a 1RM of 70.7 kg at a weight 74.8 kg for a 0.95 BW bench while the 10 set group had a 1RM of 79.7kg at a weight of 77.5 kg for just over a bodyweight bench. This is a little weird and you’d expect individuals with this much weight training experience to be stronger. There is also the 10kg/22 lb difference between groups though it’s hard to say if that’s meaningful or not. Certainly they were weaker than the subjects in Ostrowski which could impact the results.
Subjects were placed into one of two groups who performed what was called a Modified German Volume Training (GVT) program (GVT was originally popularized by Charles Poliquin and consisted of 10 sets of 10 at 60% of 1RM). The following split routine was used and body comp was measured by DEXA with muscle thickness of biceps, triceps and quad being done with Ultrasound.
|Day 1||Day 2||Day 3|
|Flat bench, incline bench, lat pulldown, seated row||Leg press, lunges, leg extension., leg curl, calf raise||Shoulder press, upright row, bicep curls, triceps pushdown|
The two groups performed either 5 or 10 sets of the first compound exercise for each muscle group (i.e. flat bench, pulldown) at 60% of 1RM (so the loads were submaximal) and this was followed by 3-4 sets for the secondary movement (i.e. incline bench, row) at 70% of 1RM so closer to limits. This makes it a little bit different than most studies I’ll look at since a lot of the volume was submaximal but, as above, that’s the GVT method.
Counting volume is a hassle here but the 5 set group did 8-9 sets hitting triceps on Day 1 (5 sets bench, 3-4 sets incline) and another 8-9 on Day 3 (5 sets press, 5 sets tricep pushdown) for a total of 16-18 sets/week while the 10 set group would have done 13-14 sets on Day 1 and another 13-14 sets on the arm day for 26-28 sets per week. Biceps is vaguely in the same range depending on if you count upright row as hitting biceps (it is a pulling movement). For legs, quads would have gotten 5 or 10 sets of leg press plus 3-4 of lunges and 3-4 of leg extensions so that’s 11-13 sets (5 + 5-8) or 16-18 sets (10 + 6-8).
Looking first at overall body composition, the results for lean body mass and regional lean body mass appear below. I’ve shown the largest changes in each in red to make it easier to see. I am not saying that these are or are not statistically significant, just looking at absolute magnitudes.
|5-set (pre)||5-set (post)||Change
||10-set (pre)||10-set (post)||Change
|Trunk LBM (kg)
|Arm LBM (kg)
|Legs LBM (kg)
The researchers reported that the 5-set group increased total LBM by 2.7% while 10 set increased by only 1.9% with the 5-set group showing a SUPERIOR response trend for trunk and arm. I think it is interesting to note that, while non significant, the leg growth was higher in absolute terms although overall small in magnitude (0.5 kg vs. 0.1 kg). Let’s face it, 5X not a lot is still not a lot.
Looking at muscle thickness, they report the following data:
|Muscle||5-set (Pre)||5-Set (post)||Change||10-set (Pre)||10-set (Post)||Change|
Looking at muscle thickness, they report no significant interactions for either group meaning that everybody grew the same and the results are kind of all over the place. I’ve put the largest changes in thickness in red above and you can see that 5-sets is better for biceps and anterior thigh and 10-sets was better for triceps and posterior thigh. So there’s really no overall advantage or no aggregate advantage for one or the other training volumes. 5 set was superior on two measures, 10 sets was superior on two measures and now we go to Thunderdome to find a winner.
I suppose it’s possible that this represents a difference in optimal volumes for biceps and triceps but it’s more likely that the small number of subjects making it underpowered to detect if this was a true difference or just statistical noise. It might also be due to whether or not you want to count upright rows as a biceps movement which would change the set counts a bit but the difference would be small. But overall neither group showed any real advantage.
Admittedly this study was small and short but overall it found that higher volume did not generate any further growth than lower in terms of actual muscle thickness. Specifically 26-28 sets per week for arms (counting compound and direct work) was not superior to 18-19 sets per week. The leg volumes were very similar, 12-13 vs. 16-18 sets which might explain the lack of differences there.
Again, leg LBM gains were a little higher for the higher volume group which fits with anecdotal beliefs that legs need more volume but the differences were still small overall. But the volumes were so close to make it irrelevant. Once again without truly higher volumes for legs, we can’t really draw that many conclusions. Perhaps 20+ sets for legs would be better but this study didn’t examine it. As well the volume for triceps and biceps was near the high end so we don’t know if lower volumes would have been equally effective. Without data, no conclusions can be made.
Overall, this study found no difference in upper body growth with very high versus lower volumes although the leg volumes were so close together it’s hard to tell. Optimal growth was found between 12-18 sets for legs and 18-19 sets for upper body with 26-28 sets showing no greater growth. Whether lower bicep/tricep volume would have been as effective or higher leg volume more effective is unknown.
Effects of a 12-Week Modified German Volume Training Program on Muscle Strength and Hypertrophy—A Pilot Study
This was a follow-up paper to the previous one done by the same group with Hackett as the primary author. It was literally identical in structure and methodology with the primary difference being that it was 12 weeks rather than 6. This was done to address the fact that the original had found marginally different results in terms of strength gains being greater in the lower volume work compared to other studies. They comment that other researchers suggest that studies need to be at least 12 weeks or more to show meaningful results and they wanted to see if their previous results were real or just an artifact of it being so short.
Towards that goal, 12 males (an admittedly small number) with a minimum of 1 year training experience were recruited and split into one of two groups with the design being identical to the previous study in terms of the workout done and the number of sets. So I won’t repeat that here. Pretty much the entire methodology was identical. Genuinely, it was just twice as long as before. They had similar levels of strength as in the previous study. The 5-set group weighed 75kg with a starting bench of 76kg or bodyweight and the 10-set group weighed 83kg with about a 82kg 1RM bench so again bodyweight bench. Fairly beginner but that’s consistent with the minimum 1 year training experience.
The real limitation to this study was the measurement of body composition and lean body mass, fat mass via DEXA with no direct measurement of muscle thickness being made. As I’ve mentioned, LBM gains are not a fantastic indicator of actual muscle gains and are, at best, kind of indicative of what’s going on. It’s a little curious that the 6-week study used Ultrasound and this one didn’t but the title of this paper…a Pilot Study suggests it might actually have been done first. Generally speaking, pilot studies are small with a reduced methodology (reducing cost) to see if it’s worth doing a longer/larger study or not. Really just a proof of concept study before you throw the big bucks at bigger studies. Which makes me wonder why this is written as if it were done second. I truly have no idea what’s going on.
Looking at body composition changes the 5-set group gained weight and LBM throughout with a small increase in fat mass while the 10-set group did not (while not statistically significant, their average weight went up in the first 6 weeks and then decreased). As per the previous study, trunk, leg and arm mass was measured and there were no differences between group (so the lower volume worked just as well as the higher). Of some interest, leg mass started to decrease from week 6 to 12 in the 10 set group although the exact reason is unclear. The researcher note that the 10-set group started to lose weight and BF% in the second half of the study so this might have been a diet issue.
|5-Set (Pre)||5-Set (6 wks)||5-Set (12 wks)||10 set (pre)||10-set (6 wks)||10 set (12 wks)|
|Trunk LBM (kg)
|Arm LBM (kg)
|Legs LBM (kg)
Of perhaps more importance, while the 5-set group gained a small amount of body fat, the 10-set group lost a small amount. This suggests that the diet might have varied with only the 5-set group being in a surplus while the 10-set group was in a deficit. This is always a problem with such studies (and why global measures of body composition are informative in addition to direct muscular size), the lack of dietary control. But it could have altered the results to be sure.
Looking at strength gains, neither group increased their leg press 1 rep max which is weird but the 5-set group is reported as being the only group showing strength gains in terms of bench 1RM. This occurred over the first 6 weeks of the study but I think this is a misrepresentation of what actually happened. I’ve shown the strength changes below.
As you can see below, the 5-set group started off with a lower 1RM bench press which wasn’t statistically significant but is probably real world significant (7ish kg = 15 ish pounds) and basically caught up over the first 6 weeks. So yeah, they made more strength gains to end up just as strong so meh. And both groups made the same small improvement from week 6 to 12 both of which were significantly different from starting levels. So I don’t think this means much. The bigger conclusion is that, over the length of the study, the absolute strength gains were really no different between groups despite the different training volumes: the moderate upper body volumes worked just as well as the higher. Once the 5-set group caught up, there was no meaningful difference in gains from that point until the end of the study. Even the slope of the line is basically identical.
I am not going to present the detailed statistical analysis they did in terms of effect sizes but only the conclusions. They state:
The trivial effects for leg lean mass in the 5-SET group compared to the small effect found for the 10-SET group at six weeks suggests that a greater volume over a relatively short duration is effective for leg hypertrophy gains. These trivial effects on leg lean mass were maintained over the whole 12 weeks for the 5-SET group, whereas, as discussed previously, there was an unusual decrease for this measure in the 10-SET group. In contrast, for the upper body hypertrophy measure at each time point, there was a tendency toward slightly greater effects for the 5-SET compared to 10-SET group. This may also point toward a different response to resistance training volumes for muscles of the upper and lower body.
Which is basically a long way of saying that statistically, the higher volume group was a little better for leg mass gains while the lower volume group was a little better for upper body mass gains. But the effects were tiny in either direction. A trivial versus small effect just doesn’t mean a lot in statistical jargon. Even the researchers didn’t think it meant much and only mentioned it for completeness. It is at least generally similar to the previous study in terms of the results and directions of the results. Anecdotally, many feel that the lower body needs more volume than the upper and this speaks to that if just barely.
Since is bears repeating, the 5 set group did 8-9 sets hitting triceps on Day 1 (5 sets bench, 3-4 sets incline) and another 8-9 on Day 3 (5 sets press, 5 sets tricep pushdown) for a total of 16-18 sets/week while the 10 set group would have done 13-14 sets on Day 1 and another 13-14 sets on the arm day for 26-28 sets per week. Biceps is vaguely in the same range depending on if you count upright row as hitting biceps. For legs, quads would have gotten 5 or 10 sets of leg press plus 3-4 of lunges and 3-4 of leg extensions so that’s 11-13 sets (5+5-8) or 16-18 sets (10 + 6-8).
Putting that together we get:
|Upper body volume||13-14 sets||26-28 sets|
|Upper body gains||Slightly better||Slightly worse|
|Lower Body volume||11-13 sets||16-18 sets|
|Lower body gains||Slightly worse||Slightly better|
With the differences being essentially insignificant in the big picture. It is interesting that the lower body did a little bit better with a rather small volume increase (only 5 sets difference) but this still doesn’t help to answer the question of whether or not higher leg volumes (i.e. 20+) would be superior. But basically the higher volumes of training didn’t outperform the lower volumes.
So overall, even doubling the training volume and workout duration (over a longer study), the results were basically the same supporting that moderate volumes are sufficient for maximal gains in the upper body. The lower volume group did just as well as the higher volume group for upper body with a small indication that legs responded to the higher (but still relatively reasonable) weekly volume and were in the same range as the moderate volume upper body set count to begin with (i.e. 11-13 or 16-18 sets vs. 13-14 sets). Since no higher lower body training volume was tested, no conclusions can be drawn as to whether that would have been superior.
And that’s where I’m going to cut Part 1 of this series. In Training Volume and Muscle Growth: Part 2, I’ll look at three newer studies in detail to see how they fit with the above before finishing up in Part 3 by trying to put together a model based on all of the data (except the Radaelli study which I will ignore going forwards).