An astrolabe is a marvellous instrument. I bought one from a Renaissance fair, and I think it's great. If you know the approximate date, you can tell the time of day when the sun is up. If you know a few stars, you can tell the time of day when the sun is down. (All assuming the sky is not too cloudy.) I have tried it several times, and it is accurate to within a few minutes.

Basic physical description: It is a metal disc with a ring at the top. The ring is for attaching a light chain or string to, so it hangs at a fixed angle.

On one side of the disc, there is an arm which rotates about the centre of the disc. There are angles engraved at the circumference of the disc, so you can tell what angle the arm is pointing at relative to horizontal (when the astrolabe is being hung from its ring, gravity provides the reference for vertical). The arm has a tab on each end that juts out from the disc. Each tab has a notch in the middle, so that if you were to draw a line between the two notches, it always passes directly over the centre of the disc (i.e. where the arm rotates). The whole purpose of this side of the astrolabe is to measure the altitude of the sun or of a star (i.e. the angle between the sun/star and the horizon). To measure the altitude of the sun: hold the astrolabe by the chain, twist it so that the sun hits the edge of the disc (rather than either of its faces). Now, rotate the arm so that the shadow of one of the notches falls on the notch on the opposite tab. The arm is now pointing directly at the sun, and the altitude can be measured by reading the engraved angle next to the arm. To measure the altitude of a star, hold the astrolabe by the chain and put it between your eye and the star. Rotate the arm until you can see the star through both notches at the same time. The altitude of the star is read the same as for the sun.

On my astrolabe, there is also a shadow square on this side of the disc (which isn't nearly as interesting as the rest of it, so I'll pass over it for now) and a calendar for converting between regular Gregorian calendar dates and a Zodiac-like calendar.

That's just one half. The other half is more interesting.

The first thing about the other side is that it has hours marked at the circumference, instead of angles. One revolution is divided into 24 equal-sized hours, and these are further subdivided by half-hour, 20-minute and 4-minute marks. The circumference is raised so that a plate sits inside it. The plate is specific to a particular latitude, and in the more expensive astrolabes it is replaceable so you can use it in most places in the world. The plate has the horizon and zenith marked on it, and almucanters - i.e., lines representing altitudes above the horizon. (The horizon itself is the zero-degrees almucanter, and zenith is the 90-degrees almucanter.) The plate does not rotate. Next is the rete. There are two retes, one for the Northern hemisphere and one for the Southern hemisphere. (You only use one at a time.) Both have the ecliptic marked on it as a ring (with a calendar on the ecliptic, marking the sun's path during the year), and several points representing stars. The rete does rotate, and it is skeletal so you can see most of the plate behind it. On top of that again is another arm, which also rotates.

If you know the date, then the point on the ecliptic of the rete with the corresponding date represents the position of the sun on the celestial sphere. If you know the altitude of the sun, and you place the point on the rete over the corresponding almucanter, then you have the rete at the correct angle for that particular time. Rotate the arm so it touches the date on the ecliptic, and the end of the arm will point to the hour of the day. If you know one of the stars on the rete, and measure the altitude of that star, then you can place the corresponding point on the rete over the corresponding almucanter, then you have the rete at the correct angle for that time again. Rotate the arm so it touches the correct date on the ecliptic again, and again it will point to the hour of the day.

It amazes me how someone got all this information figured out in such a coherent manner to make this one instrument. I spin the rete, watching the stars wheel about the North pole, imagining the slow meanderings of the sun as it gets lower in Winter and higher in Summer. One day corresponds to approximately 1 and 1/365 revolutions of the rete. (The spinning of the Earth is the main contributor to the length of the day - but even if the Earth did not spin, there would still be one day in a year as the orbit causes different parts of the Earth to point at the sun. One revolution of the rete corresponds to a sidereal day, which is not the same as a solar day.) I can easily tell you when sunrise or sunset is for any date, and from that I can tell you how many hours of daylight there will be. If you give me a time of night and a date, I can tell you which stars will be visible at that time. The dates of the equinoxes and solstices are trivially displayed.

It is an amazing and beautiful instrument.