(Unguided denoting “dumb shells,” as opposed to the Copperhead laser-guided shells.)
With an artillery spotter “walking shells in,” any artillery would become accurate given enough time and enough shots, but then the enemy knows by then what’s going on. How accurate is a howitzer these days, firing an unguided shell, on its very **first **shot?
I suspect the answer hinges largely on the training / experience / sill of the crew firing that first
Also, how accurately the spotter gave the coordinates for that shot.
Depends upon who is doing the firing and with what. There are systems that have a very high probability of a hit - within the killing zone of the shell - in just one shot.
These days a howitzer is rather more than the artillery piece, it can and often does include IT links to other systems such as gps or drone derived data, or data from other artillery pieces that have already made the shot.
Assuming an accurate grid and capable crew, the biggest two factors are firing angle and distance to target. Higher angle shots are less accurate. Accuracy also reduces with distance. Even then, you’re still dealing with probabilities. At maximum range (25km), with regular unguided munitions, you have a 50% chance of getting within about 250 meters.
How is that possible?
That’s a positioning error of less that 0.06°
How is the tube of the launcher lined up on the target with such precision?
…computers?
Don’t forget the location of the gun itself and its orientation have to be very accurate and precise as well.
I’d think practical accuracy would matter a whole lot depending on what you’re actually aiming at. If you’re shooting at infantry in the open, it probably doesn’t matter all that much, but if you’re trying to destroy a particular strongpoint or something, it would.
What’s your sense of accuracy with mortars? I watched a combined arms Marine Corps live fire exercise a few years ago, and the first shots from an 81mm mortar were way off the target. I’m guessing a range of maybe 2,000 meters? An officer made a comment to me to the effect of, “You can’t hear it from here, but over there is a sergeant who is seriously pissed.”
As for the precision of the piece itself, how repeatable the result is from one shot to the next, the recent US M777 155mm howitzer program set an ‘objective’ of 0.3% of firing range for range error probable, using unguided non rocket assisted projectiles. That means 50% of the shells should be over or under the Mean Point of Impact (MPI) in range by 60 meters or less at a range of 20km. The deflection (side to side) error probable would be a fraction of that.
As other answers suggested, the bigger picture of accuracy also considers the difference between the MPI and the point at which the crew has been told the target lies. That also depends in part on the gun, but more on how accurately it was laid, how accurate was the estimate of the target’s and gun’s position, how accurate was the weather data used to correct for temperature and wind (which could be different at the target even if exactly correct at the firing position), etc.
So loads of variables for ‘first shot’ which I think makes it unrealistic to quote a particular number for that, though there are also specifications for it under certain assumptions. But if it’s observed fire, the idea would be that skilled forward observers and crews could move the Mean Point Impact onto the target after some number of rounds, but even so any given shell would still be 50% likely to miss a point target in range by >0.3% of the firing range.
Keep in mind, the guns are registered before they are ever fired at an actual target. Kind of like zeroing a rifle, registering the guns allows the crew to make necessary adjustments to improve accuracy of future shots. So once they’re in place, they fire at some known points a few times to get everything dialed in. After that, it’s all computers.
I’m always amazed by this stuff.
Like, how Victorian era battleships were able to hit targets almost over the horizon - from a moving platform.
Last year in Afghanistan we were actually getting better accuracy from our 120mm mortars than from our artillery. There were a lot of factors at play, and a lot of “seriously pissed” people. It all got worked out eventually, though.
Well not quite that far back. Battle ranges at sea increased dramatically between Victoria’s death in 1901 and WWI. At the turn of the 19/20th century the guns would still be aimed locally individually and a couple of km was typical battle range (the USN scored a low % of hits, but enough, and the Spanish almost no hits at Santiago in 1898 from typically ~1.5km). The amazing advances in non-electronic computing and data transmission which happened in a fairly short period after that allowed naval combat ranges on the order of 20km at Jutland in 1916 (though most hits in that battle were scored at shorter ranges than that).
As a comparison to modern land artillery dispersion, in 1918 when a USN battleship squadron augmented the British Grand Fleet, some of the US ships’ salvo patterns at first proved quite wide. In a January 1918 12,000 yard ‘long range’ battle practice various British ships had patterns of 300-500 yards difference in range from longest to shortest round in a single salvo. At first only two of the US ships were in that range of precision, and others had spreads up to a 1000+ yards between longest and shortest round in a salvo. In later practices the US average came closer to the British average. But even 400 yards is 3% of range.
That was extreme spread not 50% spread, and it included aiming errors and other differences between different guns on the same ship, so isn’t directly comparable to the 0.3% range probable error of a modern gun-howitzer from shot to shot for a single gun. But it’s clear that even the seeming lack of precision of 60 meter average range miss at 20km by modern artillery with unguided rounds is still very precise compared to WWI battleships.
Thanks for the great responses (btw Bear Nenno, you used to be an artillery officer in the Army, or did I misread/misremember something from another thread?)
Another question: Does propellant reliability play much of a role? If the propellant is burning too “much” or “little” due to some internal defect, would it cause the shell to overshoot/undershoot the target?
Infantry non-commissioned officer. Still in. I have about four years left until retirement.
It must, but I don’t really know specifics on storage requirements and reliability, etc. I know that it’s possible to just get “bad ammo” and entire lots are exchanged out. Specifically what goes “bad” most often? I don’t know.
I spent about two years in a Fire Support Element in the National Guard. It’s while I was in ROTC and I was enlisted as a Simultaneous Membership Program cadet. I was not a school trained 13F (Joint Fire Support Specialist). I say again I was not a 13F but I spent a lot of time with them teaching me. I’ll reference ATP 3-09.30 Observed Fires (full pdf…that I expect nobody to read)
I never saw something like a CEP for an artillery piece. There are varying minimum safe distances based on the normal distribution of a sheaf plus IIRC some multiple of their inaccuracy. Mostly it didn’t matter because you aren’t trying to be that accurate. During adjust fire you transitione to “fire for effect” once you had your 100m bracket and split the bracket with a last 50m adjustment. That gets you within 50 meters…close enough for artillery. 
There is a type of fire mission that gives a hint to the OP question. It’s the destruction mission. Think about something like targeting a single bunker. Preferred for that type of mission is a precision munition like the Excalibur. If it’s important enough you can do it with unguided rounds. The final stage, after you’ve adjusted roughly onto target, is 10 meter adjustments. It’s never less than 10 meter adjustments. In that final stage you only adjust after firing three rounds one at a time. Like I said it’s a hint. The bunker may well be smaller than 10m. As a concept CEP only talks about 50% of rounds hitting in that circle. Like Bear_Nenno and Corey_El points out range matters. Procedure points to not trusting anything less than 10m though.
As another hint there’s the registration mission where the guns are registered to a known point. Registration pushes down to a 25 meter bracket. That mission completes when two rounds each have been fired over and short of the target. Final refinement data is then sent to narrow the bracket. That refinement data is sent to the nearest 10m increment. There’s some averaging pushed by four rounds in that final bracket that furthers the hint. If they were accurate enough to be reliable with one over and one short on the 25m bracket there wouldn’t be a need for the extra rounds.
Putting the procedures for destruction and registration missions together points towards a good chance of being able to get a single round within 50 meters of a known point on the first shot. Things like extreme range or high angle fire introduce issues. That’s good enough for something like troops in the open or light vehicles that you would be most worried about running away or seeking cover between shots. The hard part is having the point you are shooting at actually be the point where the target you want to hit is sitting.
…and an aside driven by the assumption of bracketing. Trained and experienced observers don’t necessarily have to bracket, either. It’s the teaching method because it works as long as you follow the procedure. You start by teaching the method for inexperienced troops, bracketing. It’s just a slow procedure. ATP 3-09.30 also talks about hasty bracketing and one round adjustment techniques for more experienced observers. As a tanker, I usually completely skipped over adjust fire and went straight to fire for effect on my initial call for fire. That got mentioned as a technique during artillery training at my Armor Officer Basic Course because of how fast our tactical situations could change. We got trained and tested on adjusting fire using bracketing though. That’s where you start with a class of completely inexperienced Second Lieutenants.
Anecdotal: a really good gun crew could be ridiculously precise even back in World War II. Allegedly.
The US 333rd Field Artillery in WWII famously hit a Tiger tank at over nine miles range:
A corroborating web page briefly alludes to the incident:
http://www.40daysofhonor.com/Day-1-Jimmie-Lee-Leatherwood.html
Those are the kind of anecdotes I had in mind pointing out the difference between precision of the piece itself and overall accuracy of the whole system including observation, position plotting, gun laying, weather correction and so forth.
Even if you know, or knew for WWII pieces which had similar accuracy to modern ones with unguided non-rocket assisted rounds, the exact position of the target and gun, do the calculations for gunlaying exactly right, actually lay the gun exactly, weather doesn’t vary between gun and target, etc. the piece is just not precise enough to have a significant chance of hitting a tank size target at long range on any given round. At range error probable 0.25% of range, quoted for some US WWII pieces (actually a little less than the ‘design objective’ for the recent M777 155mm), 50% of rounds would over/under by more than 40 yards at 16,000yds, even a Tiger II was only around 8 yards long and 4 yards wide.
Although obviously it’s possible. But it’s to be kept in mind in reports of WWII indirect artillery fire, land or naval, knocking out tanks. It was less often the case when reported from the tank side. OTOH artillery fire could often foil combined tank/infantry attacks by preventing infantry accompanying the tanks, and could make armor units lose their nerve and retreat.
A WWII example known from both sides is the counter attack by the Hermann Goering Panzer Div. v the US beachhead at Gela, Sicily in July 1943. US cruisers and destroyers offshore claimed to have destroyed many tanks with direct hits. The Pz Div’s report says a single Pz.III was destroyed outright by naval gunfire among 2 Pz.III’s and an armored command vehicle destroyed in total and another Pz.III immobilized and blown up to prevent capture. However, the German force did retreat and the beachhead held. ‘The planned attack on Gela was cancelled because of ammunition and fuel shortages’ according to the German report, but it acknowledged that the enemy was supported by ‘very heavy artillery fire’.
These are great points. To elaborate:
It’s not just possible to hit a tank directly with an unguided artillery shell from 16,000 yards—fire enough shells, and it becomes probable.
And of course, the number of shells fired in WWII was way more than “enough.” It would be a lot more extraordinary if we got through an entire world war and no tanks were ever hit directly from very far away.
Hitting a tank directly from nine miles away is indeed extraordinary, but extraordinary things happen all the time. But human beings are notoriously bad at intuiting probabilities, so that anecdote tends to impress people more than it should.
Just a few days ago, I heard a Super Bowl announcer refer earnestly to the “hot hand” phenomenon as though it were objectively true. That’s silly, but humans seem to be wired to see causation everywhere. It would be wrong—but very human—for that artillery crew to think that their they hit that tank due to their skill rather than luck.