I’m guessing that Roscoe’s code is an earlier stage in the evolution; note that Roscoe’s code has the 10 signal positions in a fairly logical, memorable sequence. Perhaps they changed the pattern for the Citadel because their codebook (or at least the signals they used often) had more 0’s than 6’s, or something; so they modified the signal positions to make a 0 easier and a 6 harder.

http://www.coastalradio.org.uk/spud/spud/spud02.pdf

http://www.navy.mil/navydata/questions/bells.html

The full encoding is (where every movement is followed by a return to the baseline): 0 = [left down]; 1 = [right up]; 2 = [right down]; 3 = [middle up]; 4 = [middle down]; 5 = [left up]; 6 = [right up up]; 7 = [middle up up]; 8 = [left up up]; 9 = [{left, middle, right} down down].

In this interpretation the viewer only has to see the disks, not the mast and crossarm, because every halyard has exactly one disk, all the time. Because of the offset of the middle halyard left and right are easily distinguished, and the up and down movements are also easy to distinguish because only one disk is moving up or down relative to the two other disks at the baseline. At night the disks can be replaced by lamps.

It’s a lot easier to operate than the other encodings, because you don’t have to add and remove disks all the time.

Nemo argued earlier that a distant viewer could not determine absolute vertical position; Barry argues that the viewer cannot determine absolute horizontal position. It seems to me that both of these assertions are true if the viewer can see only the disks, but both are false if the viewer can also make out the mast and the crossarm. I don’t know whether the mast and crossarm were assumed visible or not, but it seems likely that the assumption would be either both or neither.

I’m going to add a photo of the signal mast to the post.

@Nemo: Good point about relative vs. absolute position. Possibly the crossarm (yard?) on the mast is visible enough to provide a reference level. I just don’t know.

@Barry: Oops. Who am I to be telling the Royal Navy how to count?

I think you mean six, rather than five, hence 6-cubed = 216 patterns in all. For each halyard, there is one pattern with zero disks, two with one disk, and three with two disks.

But I really do not believe 8 and 9 could only differ in whether the six discs are crowded near the top or near the bottom. These would be among the harder patterns to distinguish from a distance. Imagine viewing a ship, off on the horizon… And you have to tell whether the discs are closer to the top or to the bottom of her mast?

So my proposal is that the code is unary, with N discs representing the number N-1, except that 9 discs can be 8 or 9 depending on position. With this code, you can distinguish 8 from 9 by the relative positions of the discs; i.e., are the middle discs nearer the top discs or the bottom discs?

For the code to require identifying the absolute position of any disc would seem like a design flaw…

Perhaps nearly the entire code is unary, and the difference between 8 and 9 is whether the three middle disks are closer to the top than to the bottom of the masts? I think such a code would be pretty easy to read, even from a distance, with very little chance for confusion between numbers.

If this were the code and you were assigned the task of generating a small set of mnemonic diagrams, what would your diagrams look like?

Thanks for this bit of trivia. Just googling around led me to my old ship USS Mount Whitney.

After leaving the Navy. I had a chance to visit Halifax on business.

Some trivia:
The most eastern-northern deep water port open all season.
Ammunition ship explosion during WWI?

I believe that you could use it for individual letters/numbers and code words/groups.

Fair Seas