Make that 20mm, the Phalanx round is an APDS (actually a Missile Piercing Discarding Sabot) round with a boat-tailed tungsten projectile (initially depleted uranium). Obturation and rifling engagement is accomplished with a plastic rotating band molded into an aluminum pusher plug crimped into the cartridge case. A shallow slot at the back of the projectile coins into the pusher to ensure full spin for stabilization. The reinforced nylon sabot opens from the rear and is cleared from the pusher by rotational forces and  reverse flow (overtaking) gases at muzzle exit. It has been mass produced for three decades.

Didn't know all those details, thanks for sharing.

Then again, Phalanx isn't exactly about accurate shot placement, it's about saturation.

Accuracy loss in this case may actually be a bonus, allowing for a larger beaten zone.

The original Phalanx round from a test barrel shot inside 0.1 mil at 1,000 meters. The gun and mount could easily shoot about 0..25 mils. Current ammunition is somewhat less accurate. In the late 1980s the patent owners sued the US Navy for withheld royalties. One of the Navy's defenses was that it didn't use all the patent features because they made the round too accurate and it wanted more dispersion. The argument was, and is, faulty. The Phalanx System fire-control radar tracks both the target and penetrators in flight, and uses some neat mathematical tricks to adjust the mount's aiming track during an engagement. Keep in mind that even with 3,600 rounds per minute and an 1,100 m/s muzzle velocity, penetrators in flight are almost 20 meters apart and an ASM target is moving at between 200 and 600 m/s - wider dispersion is not your friend. Of course the head-on engagement is the easiest, but for destroyers/frigates the long crossing engagement with a missile approaching an aircraft carrier is the money shot. Additionally, the Phalanx and Goalkeeper penetrators must generally penetrate the target missile warhead with enough energy to cause its detonation; the warheads are armored sufficiently to penetrate deeply into capital ships. It wasn't on my plate, but I believe the radar's discernment and accuracy is on the order of 0.1 mil (on a rolling and maneuvering ship). I presume radar tracking of projectiles in flight is still used with the optical aiming system. 

O.1mill at 1000m is like 10cm , no way any autocannon can group inside anywhere close to that. They are good if the various barrels are in sync within 0.1 mill of each other . In any case fr point for CIWS its more likely they wanted 0.5 mil to 1mil dispersion at 1000m to be able to actually hit anything.

MrT4

As I noted: 0.1 mil from a test barrel and about 0.25 mill from the mounted gun (early General Dynamics production). Not at all unusual as the system is continually correcting point of aim from the radar returns. An engagement consists of several 6-shot bursts followed by longer bursts as the picture is refined. 1 mil dispersion is almost useless on a moving target only 40 to 60 cm in diameter. 

Great information, thanks for sharing!

Indeed, I have two of those projectiles sitting less than a metre away right now.

For a GPMG is more dispersion better.... would one prefer an innaccurate GPMG... 

and would it be any different for a 556 SAW gun..? 

On the firing range, less dispersion is better, because better scores can be achieved.

In combat, the stress induced aiming error of the machine gunner totally dominates dispersion. The comparably small differences between GPMG types virtually drown in the aiming error. Because of this, and because of the very short exposure times of targets in combat, the Germans adopted fast firing GPMGs. The rate-of-fire switch of the early MG34 was dropped. 

In the GPMG heavy machine gun role (from tripod) and as a tank-machine gun, where long burst are fired, a high cyclic rate is also seen as advantageous, because tactically it is not irrelevant whether a 50 round burst takes 5 seconds or only half the time. (Or if you have 5 seconds, wheter your MG fires 50 rounds or 100 rounds at the target.)               

Right.  I heard that the BAR sometimes had complaints of being too accurate,  the users wishing for more dispersion in ww2