Try this on a hold you know well. Take your usual four fingers on it, note how it feels, then lift your index finger off and hold the same position with the other three.
It gets harder immediately, and not by a quarter. It gets harder in a way that feels different, like the hold has started trying to twist out of your hand rather than just weighing more.
The two subsets, defined
A front three is index, middle and ring, with the pinky lifted off. It's what a lot of climbers default to on narrow edges without ever deciding to, and it is the subset that shows up in three-finger pockets set toward the thumb side of the hand.
A back three is middle, ring and pinky, with the index lifted off. It appears on pockets set the other way, on slots too narrow for four fingers, and in any position where the index has nowhere useful to go.
Everyone who climbs has met both. Almost nobody has trained either on purpose, which means most people have no idea which of the two is quietly setting their limit.
Why the back three is so much harder
Two things are happening. The first is that your middle and ring fingers do not really have separate engines. The muscles that flex them share a common belly, and the small muscles inside your hand tie the fingers together further. That arrangement is fine when the whole array is working, because the fingers are cooperating anyway. Take the index away and you change the tension balance across a system that was never designed to be split up neatly, and the two fingers in the middle of your remaining three are the ones carrying the change.
Ask a room of climbers and they will all say the back three is worse. They're right, and the reason is not simply that the pinky is weak.
The second thing is about anchoring, and it is the part that explains the twisting sensation. When four fingers are on a hold, you have an outer finger on each side of the array, and between them they hold the whole thing square. The load runs down through your hand. Drop the index and you have lost the anchor on one side entirely. What is left has to resist the load trying to roll sideways across it rather than pull straight down, and your hand loses the lateral stability it had for free a moment ago.
That is why a back three does not feel like a four-finger grip minus one finger. It feels like a different job, because it is one. You're not just producing less force with fewer fingers. You're producing force and controlling a rotation at the same time, with the weaker side of your hand doing the controlling.
Compare that to a front three, where the finger you dropped was the pinky. You lose the least capable contributor and you keep the index, which is a genuinely useful anchor on the outside. The front three is meaningfully easier for structural reasons, not because your pinky was contributing nothing.
What loading it actually exposes
Here is the practical value of all this, and it is diagnostic before it is anything else.
A four-finger number is a total. It tells you what your hand produced without telling you who produced it, in exactly the way a two-hand lift hides which arm was doing the work. Subset work breaks the total apart. If your front three is close to your four-finger number and your back three is a long way behind it, you have learned something specific: a lot of what you thought was hand strength was index and middle, and the far side of your hand has been along for the ride.
That is worth knowing because the rock does not let you choose. A pocket decides which fingers go in it. If your capacity is concentrated in two fingers, every hold that excludes them is a hold you are meeting at a disadvantage you did not know you had.
Uneven strength across fingers is also the normal state of a climber's hand rather than a defect. The research cited in our post on what an unlevel edge is and why to use one documents interdigit strength asymmetries in climbers directly. The goal is not to make all your fingers equal. It is to find out how uneven you are, and whether the unevenness is costing you on the holds you actually meet.
Adding it without turning it into a stunt
The whole appeal of subset work is that it is efficient per finger, and that is also the thing that makes it easy to get wrong. Put the same load on three fingers that you were putting on four and you have handed each remaining finger a raise. Put it on two and you have done something significantly more aggressive than you probably intended.
So the arithmetic comes first: your back-three load is going to be well under your four-finger load, and if it isn't, you've either been under-loading your four-finger training or you are about to find out something unpleasant. Start lower than feels necessary and let the number tell you where it belongs.
Beyond that, we're not going to hand you a set-and-rep scheme, because we don't have one written down that we would stand behind. What we can tell you is the shape of it: subsets belong alongside your normal four-finger training as a small accessory, not instead of it, in the same way our post on the grips your hangboard cannot train treats claws and rolling edges. Log them as their own line, because a back-three number and a four-finger number are different tests and comparing them directly tells you nothing you did not already know. All the usual hygiene from our testing post applies without modification.
One general caution, and it is the only one here. Pockets and finger subsets concentrate load into fewer structures than a full-hand grip does, on the wall and in training. It is a reason to introduce this gradually rather than a reason to avoid it, but if a finger is already tender or already tweaked, this is not the week to start. Load the back three off the wall, on a day and a load you chose: a pocket on a project is a bad place to meet it for the first time.
The connected question this raises
Once you start thinking about which fingers are carrying a hold, the next question follows immediately: on a flat edge, are they even all touching it properly? Our post on why your fingers aren't the same length and your edge assumes they are takes that up in full, including what staggered geometry does about it. It is the same theme one level down, and it is the better read if your interest here is fit rather than subsets.
The bottom line
The back three is harder than the front three for two reasons that have nothing to do with the pinky being feeble: your middle and ring fingers do not have independent engines, and dropping the index costs you an outer anchor, so the load twists instead of pulling straight. Loading the subsets on purpose tells you how much of your hand strength was really two fingers, which is a thing worth knowing before a pocket tells you instead. Start lighter than the arithmetic suggests, keep it as an accessory, log it separately, and build the position somewhere you chose rather than somewhere the rock chose for you.