The Analytic Fitness™ Dictionary – Loaded Carries

This entry of the Analytic Fitness™ Dictionary looks at what should be anyone’s first choice for strength training (but rarely is): loaded carries (Around 4.400 words, estimated reading time 22 min)

Both historical precedent and up-to-date biomechanical analysis suggest that loaded carries should be a mainstay of virtually every exercise regimen.

The tale Milo of Croton testifies that loaded carries were recognized as a strength-building tool in Western antiquity, while late 20th-early 21st-century case-studies in biomechanics found that loaded carries exhibit a unique biomechanical profile yielding benefits that no other forms of resistance exercise can offer.

And yet, even if they have enjoyed a surge of popularity among sports coaches, loaded carries remain virtually absent from fitness program aimed at the general public. Reasons abound for this situation and most of them have to do with the competing business model in the fitness industry (i.e. specialty equipment with certification courses vs. group classes with certification courses vs. DVD/CD/DLC home training program with or without certification courses, etc.)

So far, no-one seems to have to come up with reasons why picking up something off the floor and carrying it for a while would require the special type of instruction that can only be delivered by a weekend certification course. But this may change sometime soon and this post is an effort to stay ahead of the fitness industry curve. In my usual manner, it combines ancient history with state-of-the-art biomechanics, so without further ado, let’s start with the ancient.

A Roman Fetish

Milo of Croton and his calf (later, his bull) are some evidence that loaded carries were recognized by Ancient Greeks and Romans as a strength-building tool.

Even tho the story of Milo is very likely a fabrication based on a comparison with Hercules and some common-sense calculations (as I argued elsewhere), it must have had a lasting influence throughout antiquity. To put it in contemporary words, the ability to carry heavy loads was then considered a hallmark of ‘functional’ strength. In fact, had the Romans been aware of biomechanics, they would not have chosen another training regimen. But I’m getting ahead myself here, so let’s look at some historical evidence first.

The rural population is better suited for arms […] for whom wielding iron, digging a fosse and carrying a burden is what they are used to from the country.

Vegetius, Epitoma rei militaris, Book I, §4

Case in point, Vegetius recommends in Book I of Epitoma Rei Militaris prioritizing recruitment from rural populations already accustomed to carrying loads and dig trenches over urban ones who aren’t (Book I, §4). Given his later suggestion to train recruits for loaded marches with up to 60 roman pounds (around 20kg or 43lbs) until they become effortless (Book I, §19), the preference for country recruits is meant to guaranteeing that they are already adapted to the military burden.

Just as the bold Roman in his national arms
Cruelly laden takes the road, and before
The enemy expects it stands in formation, having pitched camp

Vergil, Georgics, Book 3, 346-348[1]

The Epitoma… is a late compilation work (4th century CE) that I discussed in details here, and Vegetius is often more concerned with rhetorical impact than historical accuracy. As an example, when describing how the Roman soldiers carry their equipment, he prefers the poet Vergil (70 BCE-19 BCE) to historical sources. Then again, Vergil’s verses convey the perception of the Marian reforms fifty years or so after their enactment and we can get additional details from later sources confirming the accuracy of this perception.

In fact, loaded marches and castrametation were known means to get the troops in shape even before the Marian reforms. For all we know, Marius most likely took a page or two fo his former patron Q. Caecilius Metellus Numidicus, who had whipped the Numidian legions in fighting shape in 108 BCE with just that. The regimen was successful enough to turn a bunch of slackers into a fierce army who drove away Jugurtha’s forces after a full day of fighting at the battle of the Muthul (the only set-piece battle between the Romans and Numidians, which I covered the full story in this post.

So Marius brought his army to the place, since the men had nothing else to do, and ran a great canal. […] This canal, indeed, still bears the name of Marius.

Plutarch, Caius Marius, ch. 15.3

A reputable and usually accurate source, the historian Plutarch (46 CE- 120 CE), recounts how Marius ordered his troops to dig a canal while waiting for the right moment to attack the Cimbri and Teutones.[2] The exact location of the canal is a minor archaeological controversy, but we know that it ran about 20km between the Rhône river and, most likely, the modern Provence town of Fos-sur-Mer (as correctly stated in the French-language Wikipedia entry, not the Galician town of Foz, as incorrectly stated by the English-language Wikipedia entry).

Fossa Marianus [sic] (bottom right, just above the inscription on the river) from the Tabula Peutingeriana, a 13th-century map of the public works of the Roman Empire, possibly a copy of a lost Roman original from the time of Augustus (27 BCE-14 CE)
Fossa Marianus [sic] (bottom right, just above the inscription on the river) from the Tabula Peutingerianaa 13th-century map of the public works of the Roman Empire, possibly a copy of a lost Roman original from the time of Augustus (27 BCE-14 CE)

So, on the one hand, we have Virgil evoking the muli Mariani and putting on more-or-less equal footing (pun intended) loaded marches, castrametation (camp-building) and actual battle, as well as some pre-Marian reform precedent (Q. Caecilius Metellus). And on the other hand, we have Marius engaging them in public work. And now for the connection: digging a canal involves the same physical tasks as loaded marches and castrametation. The loads carried (dirt from the digging and stones for the lining) are heavier than those carried during the marches and the distances are shorter, but they add up over the course of a work day to quite a lot, which is tantamount to the kind of ‘interval’ training recommended today to improve the performance of a marching army when the load carried approach ⅓ of the soldiers’ bodyweight (see Mike Prevost’s discussion on the data on military rucking for the details).

In order to sustain that kind of workload, Roman soldiers had to consume a hefty amount of calories, with consequences for the Roman supply lines that I’ll leave for yet another installment of the Old School Strength series. With adequate nutrition and recovery, far from getting the troops too tired to fight, loaded carries proved time and again an effective means to maintain the troops in fighting shape. Not only did they build the strength and endurance needed for ancient warfare but they also helped prevent injuries and contribute to the overall resilience of the troops.

Two Case Studies

Fast-forward two thousand years and some pocket change, the Romans’ fetish for loaded carries dovetails nicely contemporary biomechanics.

Unfortunately, the systematic data about loaded carries is scarce. This entry is no place to rant about the oddities of strength-and-conditioning research, so I won’t dwell on the causes and go straight to the consequences: the appeal of loaded carries is based on theoretical arguments backed by case studies.

From the standpoint of epistemology and philosophy of science, this situation is rather fascinating, and I would be tempted to leave it for an aside if it were not for the extreme importance of loaded carries. I’ll try to keep things brief tho. So, first, let’s establish that case studies are how science establishes important facts without worrying too much about generality. Nowadays, “generality” in scientific contexts means “statistical significance”, so case studies are how science gets rid of statistics.

Paraphrasing Aristotle, the goal of science is to establish generalities, so there’s a hitch worth scratching here: methodologically, case studies are no better than anecdotes. But science is also interested in establishing causal relations, and statistical analysis is not the only one that can deliver it. And in particular, anecdotes can deliver causal stories, that statistical analysis could later confirm. So take the conclusions that follow with a grain of salt, but it does not have to be too big a grain.

Study I: Strongman Carries

I’ll consider only two studies here, the first of which is McGill, McDermott & Fenwick, Comparison of different strongman events: trunk muscle activation and lumbar spine motion, load, and stiffness, published in the Journal of Strength & Conditioning in 2009. As the title gives away, this study looked specifically at loaded carries within the context of strongman events and therefore only considered maximal weights.

Particularly interesting in the study is the comparison between the demands of loaded carries (Farmer’s Walks, 1- and 2-handed and Yoke Walk) and other strongman lifts (Log Press, Atlas Stones and Tire Flip). Relative to their relation, the authors hypothesized that the lifts would impose greater compressive forces on the spine than the carries, in particular on the lower back.

Figure 1 from McGill, McDermott & Fenwick (2009) with a 150 kg (75+75) 2-Hand Farmer's Walk (a), a 38 Kg 1-Hand Farmer's Walk (b) and a 220 kg Yoke Walk. Both the Farmer Walks and the Yoke Walk required more power than the subjects could develop with their hip extensors, requiring the upper body muscles to assist with gait.
Figure 1 from McGill, McDermott & Fenwick (2009) with a 150 kg (75+75) 2-Hand Farmer’s Walk (a), a 38 Kg 1-Hand Farmer’s Walk (b) and a 220 kg Yoke Walk. Both the Farmer Walks and the Yoke Walk required more power than the subjects could develop with their hip extensors, requiring the upper body muscles to assist with gait.

In order to test this hypothesis (and a few others specific to each lift/carry) they studied 3 competitive strongmen (not identified by name) of international, national, and local levels. The study as a whole is fascinating but I’ll stick to 3 interesting facts:

  • Interesting fact 1: the study did not validate the initial hypothesis about the relation between the lifts and the carry, as the Yoke Walk generated the highest compressive forces, not the Atlas Stones (as anticipated).
  • Interesting fact 2: in both the carries and the lifts, most of the compressive forces on the spine resulted from muscle co-contraction; for the carries alone, the average contribution of muscle contraction was 66% for the Yoke Walk, 72% for the Two-Hand Farmer’s Walk, and 80% for the One-Hand Farmer’s Walk.
  • Interesting Fact 3: there were notable individual differences, and individual lifting strategies that may sometimes be interpreted as the lifter’s technique being better or worse and sometimes not.

The compressive force from the muscle forces [in the Yoke Walk] was 8020 N, and the total compression (including the implement weight and the upper body) on the low back was more than 12 kN.

McGill, McDermott & Fenwick, 2009

Fact 1 & 2 show that heavy loaded carries are safe in spite of high joint compression forces on the spine because muscle co-contraction contributes to stiffness (mechanical stability, or stability2)and actually protects the mechanical integrity of the spine.

Fact 3 indicated that co-contraction was not only reflex but the result of strategies for lifting. In the Atlas Stone and Tire Flip events, these differences could easily be interpreted in terms of lifter’s technique being better or worse (for instance, depending on whether they ended up generating momentum with the lower back [worse] or not [better]). In the carries, the differences were in the intensity of the co-contraction and hence, in muscular control. From a biomechanical standpoint, more stiffness is better. Still, from a performance standpoint, there may be a point where too much effort directed at co-contraction might compromise performance, for instance, if the competitor gets tired and has to pause more often.

The carrying events challenged different abilities than the lifting events, suggesting that carrying would enhance traditional lifting-based strength programs.

McGill, McDermott & Fenwick, 2009

Then again, conditions for the trials did not require a trade-off between co-contraction and other performance parameters (the average yoke weight was 220kg carried over 8 meters). Given the methodological limitations of a 3-guy case study, the authors did not go for earth-shattering conclusions. The stronger one was that “carrying events challenged different abilities than the lifting events” and that this type of challenge should be incorporated in lifting based programs. Drawing stronger conclusions from a 3-guy case study would really amount to reading a crystal ball. Then again, I’m a philosopher and speculation is, in my case, an occupational hazard so there’s a little bit of an aside there.

Impossible task. One of the findings of McGill, McDermott & Fenwick was that “the very large moments required at the hip for abduction when performing the [Yoke Walk] or the single-arm [Farmer’s Walk] exceeded the strength capabilities of the hip. Yet, the task was completed.” (p.1161). The measured strength deficit was not negligible (the task required 112% more power than the subjects could generate from hip extension) and authors actually found out how their subjects managed a seemingly impossible task: the quadratus lumborum assisted with gait and provided the missing power. I covered this biomechanical oddity in this very dictionary, in the entry on Functional Movement and I won’t dwell on it for too long, but there’s something interesting there, essentially that a core (and thus, upper-body) muscle ends up assisting a function (gait) primarily performed by the lower body, as a mover. Hold that thought for a minute.

Loaded carries and loaded marches I: The Received View. I bet you remember the cut-off value of 1/3 of one’s bodyweight for loaded marches that I made such a big deal of in the in the Old School Strength (if you don’t, go to that post and scroll down to the part titled “Rucking, NATO, and the Roman Way”). In a nutshell, if the weight routinely carried is below this value, the loaded march performance can be improved by training unloaded cardio (trail running, swimming, biking). if the weight routinely carried is over this value, there are no improvement from cardiovascular (VO2max) training and loaded march performance improves with strength training with upper body strength being the determinant. Again, I’ll refer the reader to Mike Prevost’s Ruck Training Programs – Part I which I heartily recommend (Part 1 covers the science and Part 2 how to apply it). Now, the current hypothesis is that training the upper body trains the core, and thus the connection between lower and upper limbs, improving coordination and stability. In other words, greater strength makes the stride more efficient by cancelling the ‘noise’ created by the additional load and compensates for the change in the center of gravity.

Loaded carries and loaded marches II: An Alternative Hypothesis. Now, I believe that there is another candidate explanation as well: strength training of the upper body, properly done, trains the upper body muscles such as quadratus lumborum who can later assist in gait during long, heavy, loaded marches. If I am correct, a strong(wo)man type of training with an emphasis on heavy carries coupled with rucking/hiking would yield greater improvement in loaded marches performance other types of strength training, so this hypothesis is actually testable. Notice that Mike Prevost’s own rucking training program template already incorporates loaded carries as part of ‘core’ training, but I sure would like to see more of those.

Study II: Kettlebell Carries

After the plethora of mind-blowing findings of the strongmen case study, McGill & Marshall Kettlebell swing, snatch, and bottoms-up carry: back and hip muscle activation, motion, and low back loads also published in 2012, is a bit of a let-down when it comes to loaded carries. That’s too bad, because even with their small sample (7 healthy males of average age, height, and weight of 25.6, 176cm, and 82kg resp.) McGill & Marshall achieved statistical significance. However, the statistically significant conclusion they drew about loaded carries were fairly limited, to say the least.

Now, let’s not throw the baby with the bathwater: the study is primarily about Swings and Snatches and expounds some fascinating results about their similarities and differences. I’ll come back to some of those results in future posts, but that not today’s topic. Plus, for the nerds, there’s a juicy 4-pages “Method” section that pertains to both loaded carries, and swing and snatches, and covers the subjects, equipment (a Dragon Door 16kg cast-iron kettlebell), lab apparatus (electromyograms and force plates), the theoretical model of spine stiffness, and their relations, in particular how the model is used to correct the data (if that sounds strange to you and/or you still think that science progresses by falsifying hypotheses with data, please review this post and come back).

Spine, hip, and knee kinematics were similar for carrying a kettlebell in both the racked and bottoms-up positions as regular walking.

McGill & Marshall, 2012

Other than that, the results about kettlebell loaded carries per se take 3 measly paragraphs in a 12-page study among the 3-page result section. This can be explained by the fact that the findings about the two carries considered (One-Arm Rack Carry and One-Arm Bottom-Up Carry) were kind of a letdown: the spine, hip and knee kinematics were the same in kettlebell carries as in regular walking and they did not quantify spinal load.

On the other hand, McGill and Marshall’s hypothesis was that “the bottoms-up carry will create different muscle activation profiles than the carry of the kettlebell in the racked position” (p. 17). In fact, the only difference between carries was increased activation of the External Oblique in the Bottom-Up compared to the Rack Carry. “Profile” is a nicely vague term, so you can take their conclusion as either a partial vindication of the hypothesis or a partial correction (what did I say about falsification).

There was, however, increased activation of all the muscles involved in the kettlebell carries compared to regular walking but it was not spectacular: 0.1% of maximum voluntary contraction (MVC) for the Rack Carry, and 14.6 of MVC for Bottom-Up Carry [3]. And here, there’s something interesting because these figures are, unlike those of the strongmen case study, statistically significant. It basically tells us that adding 1/5 of one’s bodyweight well strapped to one’s upper body makes no difference in muscle activation during gait (although it still makes some differences in joint compression and shear).

We can draw two immediate practical conclusions from this study relative to anyone’s training.

  • Consequence 1: Unless you’re a girevik, don’t bother with Rack Carries, because you won’t get any of the benefits of, say, Farmer’s Walks, from them.
  • Consequence 2: If you still want to reinforce a gait pattern with Kettlebell Carries, always carry kettlebells bottom-up for some extra external obliques work.

Consequence 1 is conditional on your not being a girevik unless you are one of those that see value in the hardstyle of kettlebell lifting as a training modality for girevoy and routinely train both ways. A picture goes a long way; so if you’re racking like the guy on the left-hand side, you can forgo Kettlebell Rack Carries for, say, Farmer’s Walks. But ir you rack like a guy on the right-hand side, Kettlebell Rack Carries will give you some unique benefits (and if you wanna know why there’s an aside for that.

HS (left) and GS (right)

Consequence 2 is also critical for post-rehabilitation training, where the goal is not to restore local function in the erstwhile injured limb but to correct aberrant motor patterns (compensations) that may have ingrained as a consequence of the injury. In such a scenario, Kettlebell Bottom-up Carries reinforce the proper gait pattern by increasing muscle activation (aka neural output).

Girevik Rack Carry. If you’re a girevik, the mention of Dragon Door was probably already enough to raise the suspicion that some of the conclusions of the studies do not apply to your rack carry. A close look at the pictures in the study will confirm this impression. (And if you have no idea what I’m talking about, go to this post, scroll down to the section “Afterparty: Patriarchal Bullshit and the Martian Option”, and skip the part about boobs.) As a girevik, your rack position is biomechanically different from the one in the study and your gait pattern during a Rack Carry is biomechanically closer to a Zercher Carry, a Conan’s Wheel, or a Yoke walk. Indeed, your hips are pushed forward by isometric contraction of the glutes so that the weight would bear on your legs rather than on your back. Because this isometric contraction forces the extension of the legs, you can’t use your hip flexors, which in turn severely shortens the range of motion accessible from the knee joint. Hence, the only options left are: (1) straighten your rack à la “hardstyle” and walk with your usual gait pattern; or: (2) use the quadratus lumborum and more generally the upper-body musculature to lift the foot of the floors.

Conclusion: Coming Full Circle

Milo aside, loaded carries as fitness protocol was a military invention, and it was just a matter of time before they would come back with a vengeance.

Then again, vengeance is a dish best served cold and the whole “coming back with a vengeance” business took a long, long time, but loaded carries recently made a comeback in one of the largest army of today as a centerpiece of military fitness evaluation as shown by the video below.

The sprint-drag-carry event of the new U.S. Army Fitness Combat Test (ACFT) comprises two loaded carries mixed with two sprints. A 25-meter dash is followed by Backward Sled Drag (90lbs/40kg), then 25m+25m there-and-back-again lateral shuffle, and finally a Two-Arm Farmers Walk with 2 kettlebells (2*40lbs/18kg). The purpose of the sprint-drag-carry is to evaluate anaerobic capacity, so scoring at the event is based on performance time. The ACFT includes 5 other events testing everything from whole-body strength to aerobic fitness, muscular endurance, etc.

I discussed the post-WW2 U.S. approach to military fitness already in the Old School Strength series and drew some unflattering comparison with the Roman approach to military fitness (specifically, there, in the final aside of the section “Galea, Képi Blancs and Coveted Green Beret). As of today, the new U.S. Army’s fitness standards may be the closest thing there is to a modern, standardized version of the Marian Legions fitness practices, which is definitely a step in the right direction to improve the fitness of the Army personnel.

The ACFT has been designed as an evidence-base test but I doubt that they looked at the biomechanical studies on loaded carries too closely. As far support to training protocols go, they are as valuable as Milo’s story, plus the force plates and electromyograms readings: methodologically, biomechanics can so far only suggest hypotheses about the value of loaded carries, not support them. But the wealth of data from the US Army will produce in the coming years will probably change things considerably. And at one point, someone will come up with a fitness industry version of the ACFT and accompanying certification.

And when this inevitably comes to pass, I (or, if my leukemia get the better of me before it happens, my ghost) will stand on a bunch of plates piled up on a sled with a snide smirk and say: “I told you so.”


[1]^The full context is this: “Recruits should very frequently be made to carry a burden of up to [20kg] and route-march at the military step, since on arduous campaigns they have necessarily to carry their rations together with their arms. This should not be thought hard, once the habit has been gained, for there is nothing that continual practice does not render very easy. ‘Just as the bold Roman in his national arms/Cruelly laden takes the road, and before/The enemy expects it stands in formation, having pitched camp.’ ” (Epitoma… I, §19, trad. N.P. Milner). The figure of 20kg corresponds to the weight carried by C. Julius Caesar during the Gallic wars and is thought to have been between 33% and 50% lighter than the weight of equipment carried by the famous muli mariani after the 107 BCE reforms.

[2]^Marius’ legions nearly rebelled when they came to believe that Marius was investing in public work fame to rather than military glory (the only kind that they could share with him): “‘What cowardice, pray, has Marius discovered in us that he keeps us out of battle like women under lock and key? Come, let us act like freemen and ask him if he is waiting for other soldiers to fight in defense of Italy, and will use us as workmen all the time, whenever there is need of digging ditches and clearing out mud and diverting a river or two. ” Plutarch’s Lives, English Translation by Bernadotte Perrin (1920). According to Plutarch, Marius was actually delighted by this warlike disposition and placated the troops by invoking an oracle that promised victory if the battle was carried later than they would have liked. But that’s a story for another day, or for later today if you care to read Plutarch online on Perseus.

[3]^Maximum Voluntary Contraction is the typical baseline of exercise science studies. It is taken subject by subjects, and measures the strongest isometric contraction of the muscles to be tested a subject can achieve. As noted by McGill & Marshall, “dynamic contraction often exceeds static maximal values” (p. 24). Similarly, McGill, McDermott & Fenwick (2009) note that “Greater than 100% of this MVC is often seen when dynamic motion occurs.” (p. 1152). Accordingly MVC is not significant per se and its primary use is to normalize readings across participants of a study to account for individual differences in strength. There is an underlying simplifying assumption though, which is the following: (1) If the difference Διi, j in isometric strength between participant i and j is such that Διi, j=n, then difference Δδi, j in dynamic strength between them is Διi, j=n±ε; and: (2) ε can be neglected in practice. This assumption is “simplifying”, because the carry-over of isometric strength to dynamic strength is joint-angle specific, while MVC is typically measured at one joint angle.

Leave a Reply