Mechanical Computers and Artillery Control

When we think of a computer here in 2016, the device that we imagine is invariably an electronic computer. There’s a reason for this: the practicality and versatility of electronic computers is simply unmatched by mechanical devices. But this hasn’t always been true, and mechanical computing devices predate electronic computers by many centuries.

There area quite a few examples of mechanical computers, a prominent one being the very early Antikythera Mechanism, but we mustn’t reach so far back into the past, and into the realm of astrology, to find practical mechanical computers. A whole generation of these devices was used during, and for, the second World War.

A major demand for military field computing existed in the artillery. Artillery operate over much too long of ranges to be aimed by “eyeballing,” so some kind of sophisticated system was always needed to determine the direction and angle to fire in order to hit a far-off target. Early on, this computation was done by a human gunner, who early in the 20th century had a fairly complete set of equations available to calculate a firing solution based on data measured from a map. A forward observer with a good view of the target would then report back on where the shell had actually struck, providing a closed-loop feedback mechanism for the gunner to make small adjustments until they struck right on target.

Of course, saying “the gunner” here is quite generous. In practice, the crew of an artillery unit consisted of a large number of people, usually scattered across multiple locations. Of course a number of people were required simply to operate the guns, but in addition there were usually multiple individuals dedicated only to performing the various computations by hand. This arrangement was slow and difficult enough when targeting a static object on land. The situation was significantly more complicated when working with moving targets, where the calculations had to be done quickly. Even worse, the long travel time of the shell from gun to target meant that the movement of the target actually had to be anticipated and included in the calculations.

You might be thinking that this is exactly the kind of situation that calls for automation. You are quite correct. In fact, the very early 20th century, from about 1900 to 1915, was an incredibly rich time for specialized field mechanical computing devices. Let’s look at a few, and for fun, we’ll look at them in rough order of increasing complexity, which you’ll see does not exactly match the timeline on which they were invented.

A first situation, and a very common one, is that of guns on land firing on ships at sea. This generally happens in coastal defense, and is a particularly easy scenario to imagine where I live in San Francisco, as many coastal fortifications were built here to defend the bay and still stand today. If you live in the Bay Area, take a weekend drive into the Marin Headlands and you can see fortifications in good condition that would have contained exactly the system I’m about to describe.

The process of firing on approaching enemy ships began in two base-end stations. These were small observing posts, built on towers or placed on hills in order to have clear views and built very heavily to withstand artillery fire coming the other direction. The observers in two base-end stations would use sighting devices to measure the heading to a given target, and would report this (first by runner and later by electronic means) to a plotting room located elsewhere, usually in the same fortification as the guns themselves. You can easily imagine that the two headings can be used to calculate the direction and range from the guns using trigonometry, but imagine trying to do it by hand very quickly. Fortunately, you wouldn’t have to: the plotting room was equipped with a very specialized mechanical computing device called the plotting board.

plotting_board_wikimedia_commonsThe plotting board was fairly simple and fast to operate. To start, imagine the semicircular board as an enormous protractor. At the center of the flat side is a pivot off of which stick several arms which can be swung around to various angles, read at the curved edge of the board. The first thing to do is to establish the position of the target relative to the two base end stations. Imagine a triangle: two points are the base end stations, and the third point is the target out at sea. The distance between the two base end stations would have been surveyed when they were constructed and was well known (and we will sea later, already built into the plotting board). So, we want to construct the rest of the triangle using the angles reported from the base end stations, which these stations measured as the angle away between the ship and the other base end station (the base line).

First, the straight edge of the board is called the base line arm and is assumed to represent the line (and side of the triangle) between the two base end stations. Next, the primary arm is swung to the angle reported by the first base end station. We now have on the board two sides of our triangle. The next stop is to put in the third side, but this is complicated by the fact that we have just assumed the first base end station is at the center of the board, so the second base end station is somewhere else along the base line, depending on how far apart the stations are located. For this reason, there are two arms for the second base end station, called the primary and auxiliary arms. The auxiliary arm pivots at the center of the table, and is swung to the angle reported by the second base-end station. It is connected by a linkage at the end to the secondary arm, which pivots some distance to either side of the center of the table depending on the distance between the base-end stations. The linkage ensures that the secondary and auxiliary arms are parallel, so once the auxiliary arm is set to the correct angle, the secondary arm is as well, but is some distance to the side to match the real physical arrangement.

The base line, primary arm, and secondary arm now form the three sides of our triangle, and the point at which the primary and secondary arms intersect is the position of the ship. The operators would mark this position on a large sheet of paper placed on the board and then swing the arms away so that they had a clear workspace for the next step, which is calculating how to aim the guns.

The guns were usually not located in the same place as either of the base end stations. They were often somewhere in between, and farther behind so that they were better protected by the landscape. To match this, the gun arm has an adjustable pivot point that can be slid to the sides and backward and forwards from the base line to match the position of the guns currently being aimed, and this adjustment would have been performed very precisely before the table was needed. The gun arm is then swing so that it falls immediately over the previously marked target point, and the angle that the gun arm forms is read carefully from the pivot point of the gun arm. That angle, and the range read from marks along the arm, is reported to the actual gun position for firing. Of course, nothing is ever simple. The firing solution calculated by the plotting board is for a single point, referred to as the directing point, which may or may not be the position of any of the actual guns at the gun emplacement. An adjustment on each guns’ aiming mechanism would be used to handle the offset from the directing point to the actual position of that gun.

The plotting board system was used through World War I and into the early years of World War II, before being replaced with more sophisticated automation, particularly electromechanical computers. Electromechanical computers were also key to enabling effective automated aiming of anti-aircraft guns.

Aircraft move so quickly that an almost complete degree of automation was necessary to fire on them effectively. To this end, a line of much more complicated computing devices called directors or predictors were designed. Early directors were electromechanical, while later directors were essentially modern computers, and they were built both for human observers as input and to use radar. In a common scheme for human control, two people with spotting scopes (functioning basically as miniature base end stations) would adjust a complex mount for the scopes to keep the aircraft centered under a reticule in their eyepiece. The angles of the scopes were read by the director which performed essentially the same computations as the plotting board but without human intervention. This data was then fed into a firing control computer, which completed the calculations and actually aimed the guns. Especially during World War II, these devices were quite sophisticated and standard equipment for anti-aircraft emplacements. Later models used standard military mainframe computers under the AN (Army-Navy) designation, which formed the centerpiece of military automation for many years to come.

That was just a couple of applications of mechanical computers in firing control and it’s already rather long. So, let’s save the best for next week: the Norden Bombsight, a compact electromechanical analog computer that’s often considered a major part of the allied victory in World War II.

  • history/computers/artillery.txt
  • Last modified: 2020/11/16 23:46
  • (external edit)