NavList:

Bill Morris, you wrote:
"In Whipple's time, if the shade caused no more than 20 seconds deviation, it was marked K.O.1, if no more than 40 seconds, it was marked K.O.2; and if more than 40 seconds, it was rejected. This gives us some insight as to why nineteenth century writers dwelled on prismacity and steps to eliminate its effects"

And consider, too, that only some fraction of sextants were tested this way. That means that there may well have been sextants with a full minute of arc deflection arising in individual shades, possibly a minute and a half or more on average when two or three shades were combined. Even the accepted KO2 "second-rate" shades would have yielded more than a minute of arc if two happened to combine in the same direction. Hmmm... since the shades could be turned, I wonder if sextant manufacturers made any effort to see to it that the worst shades were arranged opposite to each other so that their errors could cancel each other out when used in combination...

You'll recall that this got started when we were discussing one potential benefit of longitude by lunar altitudes over longitude by lunar distances. This clearly would have been the case back in the period when longitude by some non-chronometer method was still desirable. Today, of course, it doesn't matter anyway.

And you wrote:
"I was able to get results for six out of seven shades for each of the sextants I looked at.
Bronze octant in the style of Spencer, Browning and Co., ca 1850:
Average deviation 7.0 seconds. Worst deviation 14 seconds.
Hughes and Son sextant, 1920
Average deviation 2.2 seconds. Worst deviation 4 seconds.
Tamaya sextant 1957
Average deviation 2.0 seconds. Worst deviation 6 seconds.
Freiberger sextant ca 1964
Average deviation 2.8 seconds. Worst deviation 5 seconds."

Great data. Thanks! I wouldn't be at all suprised if 2 or 3 seconds of arc is typical today. If three shades are combined, that would amount to about 0.1' on average. Nothing to worry about except in very fine work.

And you wrote:
"It seems clear that things improved between 1850 and 1920, but little progress was made after that. Six seconds is of course a tenth of a minute and may indeed account for the "strangely different results" obtained."

Well, it would certainly be part of the puzzle. One way to resolve this is to create a shade error table. Then if we're using shades A and B with known errors of 3" and 2" respectively, we can treat their errors the same as an index correction (or we could prepare an I.C. table listing the correct I.C. for each possible combination of shades, combining the normal I.C. with the individual shade errors).

And you wrote:
"To obtain index error using the sun, an eyepiece shade can be used (it is one of the reasons why they were provided) and this will eliminate errors from prismacity of the shades, though only "lunartics" will need to do so."

Sure. And when those were a common item in sextant cases, they were very useful. Of course, modern LOP navigation doesn't need that level of accuracy, so a couple of tenths of a minute of arc here or there make no real difference.

In case anyone's wondering, you can use the same trick I described last week to get the shade error for all but the densest shades. Place your sextant flat on a table, aim it at some distant target like a radio tower or a vertical edge on a distant building, and then set a spotting scope in line where the sextant's telescope would normally be. And then do a standard I.C. test with each shade in turn (if the shade is dark, find a building that catches and reflects the light at sunset). At 25x-50x, typical for small spotting scopes, you can clearly see the sorts of small angular deviations we're talking about. If you dig around (ebay!) you can find spotting scopes like this for $30 or so. And they're good for watching birds, too.

-FER

----------------------------------------------------------------
NavList message boards and member settings: www.fer3.com/NavList
Members may optionally receive posts by email.
To cancel email delivery, send a message to NoMail[at]fer3.com
----------------------------------------------------------------