Let's take two of the sizes for Pern proposed on the demographics page, circumferences of (A) 20,400 miles (B) 4,500 miles.
(A) comes from knowing Pern has a gravity of 0.9 and assuming it has the same density as Earth.
(B) comes from evidence in the books.

From the map in Dragonsdawn, thread falls at a 24° angle to the equator, NE to SW. Since Pern has a 15° axial tilt, this suggests that the Thread cloud is in a similar but not identical elliptic plane to Pern 1. This fits with the Red Star being an astronomically recent addition to the Rukbat system. The higher the latitude, the smaller should be the area covered by a Threadfall.

The map below is from DLG, it reproduces the threadfall pattern from Dragonsdawn, same numbered falls are simultaneous. Fall 4 fell smack across the Landing site and Fall 5 was the first 'double Fall'. In fact as the map makes clear all Falls are multiple according to a fixed pattern; only when the colonists had spread out did they realise this. Thread wasn't falling on them, but on the whole planet.

If you compare say Fall 4 with Fall 5 you should see the pattern: all the Falls shift 'down 2' whilst maintaining their relative position.

'broad uneven-edged ribbons, about fifty klicks [31 miles] across of ravaged land, then broader belts of grassland and forest' - Rescue Run. This makes no sense but does give us a width at around 45° latitude. Let's make a first assumption, that this is a typical width: if widths are wider at the equator, then they will be narrower at the poles.

The length of a Fall at 45° latitude (Benden Hold): 3
(A) 1,813 miles (B) 400 miles
So the Area of a Fall is at 45° about
(A) 56,500 sq miles (B) 12,500 sq miles

10-15 minutes, according to Jayge in Renegades.
If we knew anything about atmospheric density, Thread aerodynamics we could calculate the height of a Thread cloud from this.

"We've only five thousand eight hundred and fourteen more Falls to attend..." (B'nurrin, last page Red Star Rising).

That's one Fall every 3.13 days. Is this calculated from the 16 Falls in 50 days from Dragonsdawn? If so, it's not true:

"...They fall in a definitely predictable pattern; the attacks last exactly six hours. The intervals between attacks will gradually shorten over the next few Turns as the Red Star draws closer. Then, for about forty full Turns, as the Red Star swings past and around us, the attacks occur every fourteen hours, marching across our world in a time-able fashion." (F'lar, Dragonflight)

In later books there are suggestions that Threadfalls last four hours, and since they cross three times zones logic says three hours, but let's go for six hours for the moment.

From the 'First 50 Days' threadfall distribution map in Dragonsdawn, taking the Northern continent only:
Misses      2
Singles     11
Doubles     3

So for the Six Weyrs most Falls are single. Treble falls are not possible, the northern continent is too small.

Part of almost all Falls over northern Pern will occur at sea, in the mountains, the Igen desert or over the polar ice cap where Thread is no danger and can be ignored. The first 16 Falls were on average only over inhabited land for half their duration.

In addition to this rain destroys Thread and thick clouds may well do so. I don't think dragons can see through clouds.
At night according to AMC the colder air kills Thread. Night time falls are marked on the Dragonsdawn map, I suspect someone has failed to realise that some of the Falls over inhabited areas will be at night and therefore not dangerous. This is not important to the storyline, though it does give F'lar extra brownie points for realising this.

So how many Falls in a Turn was that?
We have the equivalent of 45 Turns at 14 hours per Fall: 28,080.
Halve to allow for night: 14,040.
For the northern continent 'misses' more or less cancel out 'doubles'.
This seems to match the books, Moreta for example where days without Threadfall anywhere are said to be possible but rare.

The Dragonsdawn map has one inadvertant flaw - it implies that it is showing a complete cycle, that the 17th Fall is a repeat of the 1st. In reality a 16 Fall cycle just happens to give a close harmonic with the length of a Pern day 2, so the two falls are close together.
Let's assume that Fall 17 is alongside Fall 1 but displaced by 31 miles, the width of a Threadfall. This will continue until all of Pern is covered by Thread.
The next line of Falls on the map is 10° away, which is
(A) 567 (B) 125 miles away. So it takes (A) 20 (B) 4.5 cycles to cover the whole of Pern with Thread.
At mid-fall (16 falls, 14 hours between each) that is (A) 186 (B) 42 days.
But we need to double this as Thread at night is ineffective:
(A) 372 (B) 84 days

Threadfall can be displaced by up to 5° according to DLG, that's (A) 281 (B) 62 miles. No reason is given; high winds is an obvious factor but that can't account for the whole deviation unless Thread is taking an hour or more to fall. This seems unlikely: a parachutist would take <15 minutes from 15,000 ft (at 12mph). Thread falls too fast to be seen and dodged on the ground but slowly enough to be seen and intercepted by dragons.
The effect on the ground is that we have to worry about not just the Threadfall overhead but also ones to either side. So 'Thread days' would be every (A) 19 (B) 15 days

Note that for (A) there are a lot more 'false alerts'.

The only record of successive falls I can think of is in Dragonsong. There are two Falls about six weeks apart, and then the fall that catches Menolly out comes about two weeks after that. According to our model, she was really unlucky.

Fall 17 will not be exactly displaced by 31 miles from Fall 1 of course, we were just using that model to get the frequency. The pattern will never exactly repeat, it will have to be corrected now and then just as we add leap days every fourth year, don't add them every century, etc.

Pern is not flat but round, so Thread arrives at varying angles to the surface which means the width of Threadfalls must vary. Wider falls would be less intense of course. I would suppose this will depend on latitude (narrower at the poles) but with a seasonal variation.

And finally we have the unexpected change that precipitates the crisis in Dragonquest. This seems to be a 'hiccup' in the pattern rather than a change in it.

How often do dragons fly Thread?
14 hours, 6 Weyrs, no Thread at night... 7 days.
However most Falls will be over more than one Area, so maybe every 3 days. The average Fall is a half Fall, so by the time weather is taken into account we're down to say 2 hours of Threadfall every 3 days.

How fast does a leading edge move?
This depends on the latitude, but at 45° (Benden Hold):
(A) 300 mph (B) 66 mph
Faster at lower latitudes, slower at higher. At 30° (Half Circle Hold) [By interpolation]:
(A) 340 mph (B) 75 mph

How fast do dragons fly?
30-40 knots, according to Todd Johnson in DLG. So dragons can't outfly Thread!
In fact nothing on Pern can get out of its path except by going between. Nocturnal species that hide in burrows during the day will be fine, but how do wherries and the other avian life react?

## Conclusions

From the ground it is impossible to predict Threadfall. All that could be noticed is that Threadfalls are at least two weeks apart, but the average frequency is about (B) four times a Turn. The ability of Dragonriders to appear from nowhere and destroy Thread must have seemed magical.

Footnotes
1. Well... maybe not. That was a bit of what alt.fan.pern calls 'Annescience'. [Back]
2. In case you hadn't spotted it, it is the rotation of Pern that causes a Thread cloud to 'move'. [Back]
3. A circle of latitude at 45° is 1/(2^0.5) the length of the equator. [Back]