"Simulation Study: Musketry Effectiveness" Topic

Posted on my blog is a series of three installments describing an exercise in using simulation techniques to model musketry effectiveness on the early 19th century battlefield.

The installments are:

Part 1: Background and motivation for the study – link

Part 2: Model – link

Part 3: Results – link

To validate the model, the battle of Maida is examined in Part 3.

Jon

I would add my comment that the misfires should also be weighted so that more occur progressively as the muskets get fouled and the flints wear. So a first volley would have a much lower incidence of misfires than later ones.

Also did the Maida action have some light troops deployed in skirmish order as well as in line. This as a question is valid to casualtuies figures for both attacker and defender when trying to extract volley effectiveness?

Thanks for the comment, Rapier. Misfire rates are weighted with probability of misfire increasing with each successive discharge. See process #9 on the Simulation Process Flow.

As for Maida skirmish fire against the legere assault, the accounts I recall focused on artillery fire until the first British volley. Perhaps, the British had withdrawn skirmishers back into the parent in preparation to receive the French attack and to allow the artillery to concentrate against the attacker?

As far as I know, the Legere and the British Light Infantry were both formed in line. North of the creek, voltigeurs and Corsican rangers fought in woods in skirmish order. But the main clash south of the stream seems to have been line vs. line.

At least one account of Maida suggests that the two volley story is not what happened between the Legere and the Light Infantry. One account says they traded fire. Then Kempt ordered his men to shed their blanket rolls. The front rank turned about to get help from the rear rank. French brigadier Compere thought the British were flinching from fire and ordered a bayonet charge. Blanket rolls down, the British turned and fired a close range volley that broke the French charge. There were many issues of morale that influenced things.

"Those muskets misfiring in the first volley will attempt to be made operable in time for the second volley"

In some cases. In the excitement of combat, some soldiers may not realize the musket misfired and load another charge. Thousands of rifle-muskets were found on the field at Gettysburg with multiple charges, some as many as 14.

It seems like the musket misfire rate would be negligible compared to the number of soldiers who would not fire at the enemy. Some people estimate that number to be about 75% of the soldiers that had the opportunity to do so.

Jon:
Here is an article than might add some information:

Small Arms of the Napoleonic Wars
Re-enactor Analysis of Firing Muskets
by Greg Novak [1978]
Empire, Eagles, & Lions Vol. 1 No. 24

I have read the comments on musket ranges that appeared in "Empires, Eagles, and Lions," with interest, and would like to add my two pence as a private soldier. I have been active in the world of re-enacting since 1973, a record which hundreds can easily surpass. In that time, I have been an American Soldier of the Revolution, his British counterpart, served with Major Roberts and the rangers, and am looking into the War of 1812. En route, I have had the honor of serving field pieces from 12 lbers, through 3 lbers, and up to 6 lbers and 5 1/2" howitzers. In short, I have had a typical soldiers eye view of this period.

Having served with both the British Brown Bess (Short Land Model), and the French Model 1777, 1 would like to make a few comments about them, and flintlock weapons in general. I have found that firing independently, with live rounds, a rate of fire of two rounds a minute, to be rather easy to do. This is assuming, however, that one is not subject to orders, and that one can take shortcuts such as not returning the ramrod to its pipes, but rather sticking it into the ground next to you, etc.

Volley fire may drop this to three rounds per two minutes, depending on what orders are given. From practice, this has been the average speed that a group can load and fire for some time without any drop. The Prussian Army has been stated to have trained its troops to fire five times a minute, but I have not seen a date on this, nor have I seen whether or not it was live fire or blanks. in their drive toward the maximum number of rounds per minute, the Prussian Army adopted in the post Seven Years War period, a "self-priming touchhole," as well as a double ended ramrod. I wonder if the five rounds per minute was done with this weapon.

As for the individual performance of the weapon itself, I would like to offer the following facts, with the reminder that these weapons that I am talking about, though exact reproductions are undoubtably better made than the originals, and that the black powder used is better than that used during the Napoleonic war.

For either a Bess or a Charleville, the mis-fire rate is about one in five today. I have seen competitions run to see who could keep their musket firing the longest, when after twenty rounds, twenty-three people out of the forty who entered, were still left and the only item you were allowed to wipe the frizzen with was your thumb.

On the other extreme, I have seen a platoon of fifteen musketmen, after a battle in which they fired at the most ten rounds, attempt a "fire of joy," and not have s single shot go off.

I have seen battles fought in the rain, where the officers were attempting to explain to the sponsors that we could not fight in the rain, with the roar of musketry going on at the saw time. It takes a very heavy rain to put a musket out of operation. However, one should assume that at least one out of four weapons is in operable during a Napoleonic battle.

One additional item that rarely is considered, is the size of the ball that the musket is firing. The larger the ball, the better the accuracy. The smaller the ball, the less accurate. However, with every shot, the discharge that builds up in the barrel of a musket is impressive. in firing a .75 Brown Bess, X use a .735 ball for the first six to ten shots. By the eighth round, the barrel is down to a .74 or less, and loading the .735 ball becomes a task. At this point, I switch to a .715 ball. Considering the improvements in gunpowder, the problem must have been far worse in the 1800's.

In fact, for the first few rounds, one does not need a ram round, but rather one can pour the powder down the barrel, drop the ball on top of it, and rap the butt of the musket against the ground to seat it, thus never having to use a ramrod. After the first two shots or so, this method does not work, as the ball is liable to hang up in the barrel and not drop to the bottom, with interesting results.

This crud that builds up in the barrel, in the pan and on the frizzen and hammer, must be cleaned out. If it should get wet and dry, the resulting cake will foul up the touch hole so that the weapon will be useless. Surprisingly, the Charleville, due to its banded construction, is a far easier weapon to clean in the field than the Brown Bess. A Charleville can be field cleaned, with the help of a screwdriver in less than an hour, while to do the same to a Bess, takes a hammer and a pin driver, and more than an hour. There is also the danger, that if one losses the pins, too many times, they will not hold.

A second area where the Charleville has an interesting edge on the Bess, is in the nature of the hammer. The Bess has a goose neck, which can shatter, while the Charleville has a double neck hanvaer. Also, to adjust the flint in a Bess, one needs a screwdriver of somie sort, while the Charleville has a hole through which one can use a nail or something similar to tighten or loosen the flint. It is a small matter, to say the least, but to the man in the field, trying to adjust a flint to get his weapon to work, small things like this become important.

As far as ranges go, seventy-five yards is about the limit for aimed fire at a human sized target for the average shot. Volley fire is accurate to 250 yards still has in this sense is a relative term. More correctly, volley fire at 250 yards still has some effect, for in shoots we found about ten percent accuracy at this range, shooting at human stillettes. (There were five members in the firing party, with four volleys fired).

Closing the range, we went up thirty-three percent at 150 yards, and to sixty-six percent at 50 yards. This includes multiple hits on the target, no attempt at allowing misfires, etc.

One point on musketry, that I feel is very important, though often missed is that the first volley is usually the best, and has the fewest number of misfires. This is due to the fact that this is the round that is loaded usually without hurry, and the firer has had a chance to check his lock. The flint has not yet chipped, the frizzen is still clean, and in short, this is the volley that has the best chance of succeeding in doing the most damage.

In short, at least between the Bess and the Charleville, there is not much of a difference, although there are those who favor the French Charleville. Musket fire should have some effect on forme- d lines at about 250 yards, with the effect raising as one closes. the range. My personal house rules allow for the following losses per firing figure, (without any modifiers- these figures are for one line versus another in an open field).

Ranges
0-100 Yards 100-175 175-250
Losses
Max. 50% 25% 12.5%
Average 35% 17% 8%
Min. 20% 10% 5%

******************************

Bill H.

Jon:

Thanks for sharing your study. That was a lot of work. Have you thought of taking if further? At this point you have used three ordinance trials and one event and tested it against one event [or validated it]

More ordinance tests from the period and a lot more event tests would be necessary before you could actually trust your conclusions.

Not because your math or methods are wrong, but because:

1. You have too many variables in any one event you might chose… For instance, the artillery could have done more or less damage, or the Battle of Gohrde event. Where those 66 Frenchmen aiming or simply 'presenting arms' and volleying?

2. You have used conclusions from secondary sources on rates, events, casualties and such, which have to be checked themselves.

3. Your test is a specific tactic, which had a fairly narrow event window. Quite reasonable to use it, but it tests only one band of your entire range of effects. For example, it would need to be tested against volley fire starting from longer ranges, such as 150 and 250 to see if your conclusion that fewer volleys at closer ranges have more effect is correct. Then again, what about longer firefights between two opponents, such as Albuera. Will your calculations hold up then?

To move from the plausible to confirmed results, where you can trust the results to mirror reality, you would need to compare your calculations to at least thirty different events or more. That is possible… such information is available. With that size base, the variables would reveal themselves as significant OR have little or no significance whatsoever.

Any simulation mechanics based on your calculations would need that kind of validation to be trusted to do the job. Your calculations in and of themselves isn't quite a simulation, but the data a simulation is created from.

Best Regards,

Bill H.

I use casualty figures to mark dead units when playing Napoleonics. It makes it easier to keep track of victory points and gives viewers some idea of where the action has been happening.

I got some figures done for my Russian cavalry. My pal Krisztian did the conversion and the painting and I think they look rather good. You can see some pics here.

link

Hmmm. I didn't write the above….

Where is the simulation model itself, and what was it written in?

The choice of probability distributions seems odd to me, but then I suspect that you are using them in a different way from what I am accustomed to. I would have modelled each individual firing musket as a renewal process, and let the modest variation in reload times produce progressively more ragged volleys as time goes on -- which raises the question of why volley fire was preferred, what effect simultaneity of shooting was considered to have, and how to model that.

The attempt at validation against an historical result is a beautiful example of why this is practice is considerably less useful than beginners think it is. We have a curve showing the spread of model results. An observation from an historical battle lies somehwere beneath this curve. Is this validation? Not much, from a single observation, and history doesn't give you many numerical observations you can trust.

The result showing the disbenefit of opening fire at too long a range is, as far as I know, original in attributing the effect to the decline in musket performance as firing progresses, and seems to merit further study. I had always been led to believe that the effect was accounted for by the deeply-curved trajectories of low-velocity ball and the fact that sights set for, say, 300 yards were not adjusted as the enemy closed, thus putting most of the close-in shots over the enemy's heads. But this may well be tosh, and I might need to dig out my GB drag curve program and see how far this explanation is believable.

A most interesting piece of work -- many thanks for sharing it, I hope you'll be prepared to expose more of its inner workings (which is the way validation really happeans).

All the best,

John.

Jon (Bill & John),

Knowing that I love this particular topic, a good friend of mine (Ned Z.) made me aware of your recent musketry modeling effort and the present discussion.

After reading through it, I think you've conducted some impeccable analysis in this area so far.

I have no real criticism of what you've done or why. I'm more interested in exchanging ideas and then consider how we might collectively build upon some of these ideas after the discussion.

Let me provide you with just a little background information about some analysis I've done in the same area over the last couple of years, which I think you can appreciate and may even pique your interest.

The first and most significant results from my preliminary analysis of historical musket performance is that Nafziger and some of the other historians working in this area have done us a considerable disservice by considering the Scharnhorst 1810 Musket Trials as just six (6) discrete data points within each of the four (4) range bands he investigated.

What this singular approach ignores is the fact that each of these test firings, for each type of musket, at each range band was conducted with 200 separate and thus, independent musket shots. This enables you to consider each of these musket firings as a separate Bernoulli test trial.

In the theory of probability and statistics, a Bernoulli test trial is an experiment whose outcome is random and can be either of two (2) possible outcomes: "success" or "failure".

A Bernoulli test trial (i.e. binomial experiment) possesses the following properties:
1. The experiment consists of n identical trials.
2. Each trial results in one of two outcomes; the outcomes are often called a "Success" (S) and a "Failure" (F) or ‘Hit' & ‘Miss', which would be the case in these Scharnhorst tests.
3. The probability of a success on a single trial is equal to p and remains the same from trial to trial. The probability of a failure is q = 1 – p.
4. The trials are independent.
5. The random variable of interest is Y, the number of successes observed during the n trials.

Thus, in respects to the Scharnhorst musket trials we are actually talking about is a total of 4,800 musket data points (a statistically valid sample size)…Not, just the ‘24 total test trials' that most historians have always focused upon in the past.

So, given what I've stated above makes sense to you, is it compelling enough for you to consider re-running your model using data from a Binomial Distribution instead of the Beta Distribution you used during the original excursion?

I have no issue with the ‘level of uncertainty' that a Beta Distribution can allow you to mitigate. I'm suggesting adjusting your work in this way would reduce the historical validation effort because it is now based upon a historically & statistically significant sample size of 4,800 data points instead of an insignificant sample size (n<30) of 24 data points.

What do you think of that?

Regards,

James

I'm suggesting adjusting your work in this way would reduce the historical validation effort because it is now based upon a historically & statistically significant sample size of 4,800 data points instead of an insignificant sample size (n<30) of 24 data points.

What do you think of that?

Hey James:
I agree with your analysis concerning the weapon accuracy/performance. The limit is the context. Misfires aren't counted, nor fouling. So while accuracy could be set, the numbers fired would still be up for grabs. Regardless, the actual calculated results have to be tested against known events to establish overall validity of the calculations. As you point out… a low base [n] isn't adequate, though an [n= or >30] is significant for events comparison, depending on the results…

Regards,

Bill

Bill,

Thanks! I was a little concerned that the 'main point' (i.e. 4,800 valid data points) would be lost, as a result of the need to show some kind of proof.

It's the first of a couple of discussion points I was hoping would interest you all.

The next one is a discussion over the dynamics of the two (2) fundamental Napoleonic firing philosophies; 'Big Bang' & 'Steady State'.

However, I don't want to hijack this discussion or go off on some tangent until after everyone has had an opportunity to chime in on this point.

Regards,

James

I wonder if a bed sheet target is not the best way to count hits. I would think individual human sized wooden targets would be a better way to determine hit distribution. That way multiple hits on a single target would only count as 1 hit.
The test only deals with 2 rank lines. How does a 3 rank line replace losses in the first two ranks? Or do they just leave the missing spots open?

Mobius:

Most period ordinance tests used a battalion front-sized target [in column]. You have raised one of the many questions concerning variables. Third ranks would step into the empty space. Leaving holes didn't help with unit cohesion.

some of the many variables would be:

Smoke.
Casualties if being fired at.
Cannon hits vs musket on the same target.
Three rank vs two ranks.
Training.
Expectations of a long or short firefight.
How long controlled volleys would be used before devolving to 'battle fire.'
weapon fouling.
Weapon malfunction.
loading errors.
Weapon quality.
Terrain.
Leadership
Weather
etc…

What this means is that in the end, it is difficult to correlate final results to calculated/predicted hits. That is why it is important to create a large base of event results where the variables can be weeded out or prove to be insignificant.

Mobius says,

I wonder if a bed sheet target is not the best way to count hits. I would think individual human sized wooden targets would be a better way to determine hit distribution. That way multiple hits on a single target would only count as 1 hit.
The test only deals with 2 rank lines. How does a 3 rank line replace losses in the first two ranks? Or do they just leave the missing spots open?

My study did adjust for multiple hits on the same target and the target was treated as three-rank at the start. As casualties were suffered from the front rank, second and third rankers stepped forward to fill their position. Only after the attacker was reduced to a single rank were casualties adjusted for the probability of hitting a gap.

Jon

Pictors Studios, says,

It seems like the musket misfire rate would be negligible compared to the number of soldiers who would not fire at the enemy. Some people estimate that number to be about 75% of the soldiers that had the opportunity to do so.

That may well be true as several readers of the blog noted. Not firing, whether due to weapon malfunction or choice could be modelled under the same misfire rate variable. If you assert that 75% of firing muskets choose not to fire during any given volley, then simply increase the misfire rate and see what happens!

Jon

McLaddie says,

More ordinance tests from the period and a lot more event tests would be necessary before you could actually trust your conclusions.

My study was an exercise only. No claim was made that conclusions drawn from the exercise should be treated as anything more than that. Did my study have the rigor required of a peer-reviewed academic journal. Probably not since that was not the motivation for this exercise.

Any simulation mechanics based on your calculations would need that kind of validation to be trusted to do the job. Your calculations in and of themselves isn't quite a simulation, but the data a simulation is created from.

The main thrust of my study was an attempt to gain insight into the limited historical record of casualty rates during the smoothbore musket era. With casualty rates as shown in musketry trials, an equal-sized, steady defender could completely annihilate an attacker. In fact, theoretical calculations suggest an attacker could be eliminated many times over before reaching the defender.

Yes, my study examined only one specific tactic. That is, a theoretical advance by a stolid attacker against an equally stolid defender. If there are additional historical scenarios to consider, I would be interested in seeing those.

I don't follow the assertion that this effort was not a simulation. Data points were gathered (albeit limited), a model built that (in my judgment only) fit the existing evidence satisfactorily, a set of situational studies were stated, effect of each scenario simulated using 10,000 repeated trials (having the input parameters and assumptions noted), and then results aggregated to produce my own inferences.

Thanks for sharing Novak's antecdotal results and game rule amendments. Overlaying Novak's data onto the study data, both Nafziger II and Freitag II casualty rate curves fit Novak's upper bound quite closely.

If this conversation is to continue, how about continuing the dialog on my blog where it is easier to respond to individual responses?

Appreciate your comments.
Jon

My study was an exercise only. No claim was made that conclusions drawn from the exercise should be treated as anything more than that. Did my study have the rigor required of a peer-reviewed academic journal. Probably not since that was not the motivation for this exercise.

Jon:
I realize that, and neither James or I were thinking about academic journal 'rigor' or criticizing what you'd done, but rather how to take the exercise to a point where you could say "This does simulate" with a comfortable degree of certainty.

You built an interesting carborator with a view to building an entire engine. We were just talking about how to get to the complete machine.

The main thrust of my study was an attempt to gain insight into the limited historical record of casualty rates during the smoothbore musket era. With casualty rates as shown in musketry trials, an equal-sized, steady defender could completely annihilate an attacker. In fact, theoretical calculations suggest an attacker could be eliminated many times over before reaching the defender.

We were talking about how to get past 'suggest', 'theoretical' and 'attempt' and closer to certainty. And as we have innumerable examples of equal-sized units not being completely annihilated in stand-up firefights and attacks such as Albuera and Salamaca, we have to take your calculations, 'if correct', and ask why?

Yes, my study examined only one specific tactic. That is, a theoretical advance by a stolid attacker against an equally stolid defender. If there are additional historical scenarios to consider, I would be interested in seeing those.

Yes, there are many. Often folks on the TMP will say that there are very limited examples or none at all, which simply isn't the case. It just takes some serious digging.

I don't follow the assertion that this effort was not a simulation. Data points were gathered (albeit limited), a model built that (in my judgment only) fit the existing evidence satisfactorily, a set of situational studies were stated, effect of each scenario simulated using 10,000 repeated trials (having the input parameters and assumptions noted), and then results aggregated to produce my own inferences.

Understood. It may be a 'patato', 'patotoe' kind of thing. Is it an engine if it can't turn-over and run? You have built a set of calulations that suggest some things with one 'test' which was off by at least 12% with guesses about actual cannon and melee casualties.

This isn't a criticism, it is just saying that you've gone fifty yards to a simulation touchdown.

Thanks for sharing Novak's antecdotal results and game rule amendments. Overlaying Novak's data onto the study data, both Nafziger II and Freitag II casualty rate curves fit Novak's upper bound quite closely.

That's what I thought too.

If this conversation is to continue, how about continuing the dialog on my blog where it is easier to respond to individual responses?

We can.

I once tried making a set of Napoleonic and Franco-Prussian rules with more detail and realism than Empire II. My focus on the line firing was that casualties reduced the firepower of a 2 rank line 1 for 1 since every man is firing. But it reduces by only 2 for 3 in a three rank line as a casualty to a firing man can be replaced by a non-firing man in the third rank. I ran a program to make a prorated table of this but it was too complicated to play in a short time. Basically, to win over players from other games it had to take less time.

Analsim,
Comments on Bernoulli vs Beta on blog.

link

Jon,

I read your comments on the 'blog'. However, it seems that I couldn't set up a blog account that would allow me to access it from my work account(i.e. US Army, No Blogs).

Regardless, I provided my response below.

As a result of my desire to be short and concise with that first message on TMP, it seems that You may not have fully appreciated the significance of what I was presenting to you about Bernoulli trials or the benefits of using them.

Let me try to clear that up a bit more right now.

Yes, the overall total is 4,800 data points. However, the breakout of these 4,800 data points would be in the form of 24x Trials, each conducted with 200x muskets.

These 24 trials would consist of four (4) tests made at each the range bands of 80m, 160m, 240m & 320 meters. Each test would consists of firing just one (1) of the six (6) different types of muskets at a time. Thus, each type of musket would fire 200 shots collectively, via 200 Soldiers aiming and pulling the triggers more or less simultaneously.

Each one of these '200 shot test trials', involving just one (1) musket (i.e. making them identical) fired by 200 'independent' Soldiers meets the property qualification to be consider a binomial experiment made up of 200 Bernoulli trials.

What this does, is provide you with a historical valid sample of each musket's 'individual hit performance' at each of the four (4) range bands.

You can then use these actual test statistics as parameters for input into a Binomial distribution that can be used to fuel and generate "historically valid Hit data" for your simulation.

As I stated on TMP, I have no issue with the manner in which you've used the Beta distribution within your simulation.

Hopefully, you can see that Bill H. and I consider ourselves to be one of your colleagues, not one of your nemesis. As such, we share the same common interest in developing and using analytical tools that will help us gain a better understanding of these historical topics.

As one of your colleagues, I wanted you to consider using this Binomial approach because I believe that you won't be able to claim any kind of historical validity of your 'musket fire model' unless you can show that your claim is based upon historical data made that meets the recommended minimum significant sample size requirement.

The recommended minimum statistical sample size is 30. This is because, the statistical data analysis employed seeks to tests the hypotheses that the relationship between 2 or more variables may or may not be correlated (statistically significant).

The common number of 30 is also used as a minimum because most non-normal distributions met in practice are asymptotically normal at 30. It only stands to reason that the larger the sample size (i.e. 200 Bernoulli trials), the more accurate the inferences that come from a study using it.

Best Regards,

James (former US Army Weapon System Test and Experimentation Officer)

Mobius wrote:

I ran a program to make a prorated table of this but it was too complicated to play in a short time. Basically, to win over players from other games it had to take less time.

Using a complex engine to generate results does not automatically require a complex set of game routines to accomplish the same results. There are a number of methods for simplifying the process without disturbing the results…

And don't forget: A prorated table etc. only gives you one result, such as casualties. The relationship between casualties and combat results are nowhere a 1:1 relationship, so even with a solid process, does it need to be carried out each combat for a combat result?

Bill H.

Jon,

You might be interested in this book available at:
link

A SERIES OF MILITARY EXPERIMENTS OF ATTACK AND DEFENCE 1806

"Attack and defence experiments made in Hyde Park in 1802, by order of the Duke of York, C-in-C of the `Army, to find out how long attacks took, and how often defenders could fire at attackers early in the Napoleonic Wars..
The Commander-in-Chief of the British Army in 1806 was the Duke of York. His innovations as described in this valuable book and practised in London's Hyde Park, were essentially manoeuvre, and the experiments consisted of trying various methods of combining infantry, artillery and cavalry on the parade ground, to evaluate their possible use in battle. At this time, early in the Napoleonic Wars, troops had to move en masse, for there was no other way that infantry in particular would be able to give effective fire on the battlefield. The accompanying artillery and cavalry, although having greater fire or shock effect, were transitory, and, then as now, only the infantry could take and hold ground. The experiments are graphically supported by a frontispiece which shows the effect of fire at various ranges, and which clearly illustrates the problems facing infantry commanders when ordering their men to fire on the enemy. The experiments here described were to find out how long various infantry and cavalry charges would take to close upon the attacked force, and to how many volleys of musket fire the advancing troops would be exposed. The results of similar experiments in Jersey, and the whole book is completed with a number of observations made by suitable commentators of the period."

– Ned Zuparko

Jon, some dumb ideas …….
Now that you have developed a baseline plot, it should be possible to modify the parameters to produce similar plots for different combat situations. This could be valuable for wargames rules.
For example, I have seen old sets of rules where different firing tactics were available to the player e.g. platoon fire, fire by rank, rapid fire and so on. Also the difference effectiveness of different types of muskets was modelled. You now have a model where you can produce plots for these variations.

I believe that the human factor element has a greater impact on combat than weapon effectiveness. So the real interest might be trying things like identifying how a unit that can fire 3 shots a minute can fare against a unit that can fire 4. Does it have a chance or not and what is that chance?
Or use the model to identify the effect of first fire.
One of the major problems in rules is that in a turn/move, several volleys could be got off, yet rules do not calculate the effects of each round of fire. The soldiers at the time felt who fired when and at what distance was a major factor in combat. Read an historical account and it will usually state who fired when and what the results were.
With a model that can determine casulaties per volley rules could actually get into this type of combat which might be quite exciting. This would be the SAGA equivalent for the horse and musket period. I think 28mm players might be interested in such a game.
You could work out if first fire as important or as unimportant as different armies believed.
Or even what would happen if the first rank kneels and the second/third rank stands.
Or is rapid fire at medium or long range worthwhile if all that happens is fouled muskets?

Ned,
Thanks for the book recommendation. I placed an order for one.
Jon

What are the muzzle velocities on these muskets?

From:
The American military library, or, Compendium of the modern …, Volume 1
By William Duane
Available at GoogleBooks:
link

Pages 203-204

"But from actual experiments it has been found that common balls with the ordinary charge, traverse 180 toises, or about 1080 feet in a second, and with stronger charges still more ; and experiment has also proved that the initial velocity, or the velocity with which the ball flies when it first leaves the muzzle of the gun, is equal to 1600 feet in a second.

A series of experiments performed to ascertain this velocity, given in a concise and familiar way may be useful to the young student, more enlarged views would require mathematical knowledge and language.

A musquet of three feet six inches barrel, carrying a ball of nearly an ounce weight, was provided, and it was proposed to try the effects of the various state of the atmosphere upon its velocity; the following is the result of the experiments.

EXPERIMENTS

Charge of powder Very moist Temperate Very dry
seven drachms. weather. atmostphere. weather.

Common war powder, 1392 feet 1542 1618

Fine do. 1569 1736 1829

Fowling powder, 1566 1703 1784

Fire work powder, 1566 1706 1779

It was observed in these experiments, that when the atmosphere was loaded with vapour, the bore of the gun was so moist five minutes after the explosion, that the tow used in cleaning it, was covered with saline particles ; but on the dry day no moisture could be perceived after an interval of several minutes, and the tow came out unsoiled.

The result of the experiment on velocity was, that in very dry weather, the velocity was greater by one seventh than when the atmosphere was loaded with vapour ; and that in relation to the powder, the fine war powder was about four per cent better.
The following experiments with guns of different calibre, further illustrate the proportions of initial velocity.

Kind of guns. Kind of powd. Wt. of charge. InitialVelocity.

A musketoon
1 ft. 10 in. Fine war, 7 drachms. 1390 ft.
barrel
Fowling, 7 1367

Fire work, 7 1372

A rifleCarabine Fine war, 7 1956

Fowling, 7 1920

Fire work, 7 1934

A firelock Fine war powder7 1736
3 feet 6 inches

The advantage of well manufactured powder is to be seen in these experiments ; and the principal object, here is answered, which by directing the attention of the student to rational causes, tend to remove the effects of common prejudices, and excite him to further enquiries on a subject, which it is the duty of every person who engages in military concerns to be acquainted with."

My apologies for the formatting, I'm not sure how to fix it.

– Ned

The Brown Bess seems to range from 1000-1200 ft./s. Is that one used in any of the tests?

Mobius,

The book does not specifically name the Brown Bess as being used, it only mentions the units the soldiers came from. For example, in one experiment it was a soldier of the 1st Battalion of the 83rd Regiment, in another they were men from the light company of the 2nd Battalion of the 18th Foot, in a third it was a flugel man of the 2nd Battalion of the 58th Regiment.

– Ned

Bill (& Jon),

I appreciate Jon's invitation and motives for wanting us to share our comments about his musket study on his blog. To that end, I acknowledge and respect the work that he has done with it and even made a couple of suggestions for him to consider, that I thought could possibly help strengthen its historical foundation. Be that as it may.

As I said above, Napoleonic Musket Effectiveness is a favorite topic of mine. And given the limited amount of historical data available to all of us, it leaves us little choice other than using statistical analysis to help gain additional insight into this area.

However, it probably goes without saying that quite a few of us on TMP (and many more lurking in the shadows) have ¡¥crunched¡¦ these very same musket effectiveness numbers themselves. And because of this fact, I for one am just as interested in discussing the methodologies and approaches that were used in these studies, as I am looking at just the results. Because we can all learn from each other that way.

Since we are talking about History and not ¡¥Tiddly Winks¡¦, the single thing that¡¦s going to separate the ¡¥men from the boys¡¦ in any of these analysis is going to be the study¡¦s ability to demonstrate that its analysis has maintained some level of Historical fidelity via some form of quantifiable Verification & Validation process.

V & V is actually a Good Thing!

Because once you try to verify and validate your results, you¡¦ll quickly realize that you can¡¦t work around the need for having validated historical data to begin with. This validated data will also tend to restrict/limit the type of approaches you can use too. Additionally, you have to ensure that you are consistent and have a solid audit trail which is traceable throughout the analysis.

Here¡¦s an example of what I¡¦m talking about to help make this point a little clearer. I¡¦m analyzing Historical Musket Effectiveness on the Napoleonic Battlefield.

Here¡¦s my basic study approach (sequentially):
1. Musket Fire„³2. Hits„³3. Casualties„³4. Effect on Unit Cohesion„³5. Close Ranks„³ 6. Carry On? ***Note Cycle Time*** „³7. Evaluate Obscurants„³8. Return to Step #1.

I know that I can use Scharnhorst¡¦s Musket test trials to historically validate Steps #1 and #2, However, not for step #3 and beyond. It stops there because Scharnhorst¡¦s tests only produced: 1) Valid ¡¥Musket Fire¡¦ and 2) Valid test ¡¥Hits¡¦. Not Casualties, or Cohesion Loss and all the rest.

Yes, you can start enumerating the factors and trade-offs that will allow you to arrive at an approximation for ¡¥Hits to Casualties¡¦, but you sacrifice a little more validity as you move away from the ¡¥pure historical form¡¦ of the original data. Make sense?

So, what do I do now?

Well? That¡¦s exactly the type of questions that I was hoping we would all eventually get around to discussing on this thread.

Regards,

James