Weeks may well be the first truly man-made measure of time. The year is based on the revolution of the earth around the around sun, the month was originally based on the phases of the moon, and the day, obviously, is about the rotation of the sun. Where do we get weeks?
Weeks were designed around market days. Market days were traditionally on the same day of the cycle, but not necessarily the same day of the cycle in every town, if that makes sense. To give an example, in Turkey there are traditionally two types of markets: çarşı, which are open every day, and pazar (same root as "bazaar"), which are once a week affairs. The Grand Bazaar and the Spice Bazaar, undeniably the best known markets in Istanbul, are actually çarşıs, not pazars. The Grand Bazaar is the Kapalı Çarşı, the covered çarşı, and the Spice Bazaar is the Mısır Çarşısı, the Egyptian çarşı. There are pazars, though, on days. The most well known among Istanbullus is probably the Salı Pazarı (Tuesday Pazar) is in the hip neighborhood of Kadiköy (ancient Chalcedon) on the Asian side. My friends and I used to go to the Tarlabaşı pazarı, which was on Sundays in the down-and-ought Tarlabaşı neighborhood, near Taksim Square (the modern heart of Istanbul). I used to work in the neighborhood of Bakırköy, also on the European Side but several miles from Taksim, and my friends took me to the largest pazar around there, held in the Yeşilköy neighborhood every Wednesday. And so forth. Unlike merchants at a çarşı, merchants at a pazar would travel around to increase their exposure to new markets. You went to the çarşı, but to some degree the pazar comes to you. That should give you an idea of how "market days" worked historically, and still work to this day in some areas. Maybe where you live there are several seasonal farmers markets, and sellers rotate between where they sell.
The gap between market days hasn't been constant throughout time and space. Sociologist Eviatar Zerubavel (let's just pause for a moment to notice that his last name means "Seed of Babylon" in Hebrew) has a book called the Seven Day Circle: The History and Meaning of the Week which, in addition to investigating our seven-day week, which looks at a variety of other market cycles. For instance, he describes the various common week lengths in Africa:
Quite popular in Rwanda, Tanzania, Cameroon, Togo, and Zaire only a few decades ago, three-day, five-day, six-day, nine-day, and ten-day market cycles still regulate the economic life of various tribes in Ghana, Nigeria, and the Upper Volta. The most popular of all indigenous African weeks, however, is the four-day market week, along with the eight-day and sixteen-day weekly market cycles that have most probably derived from it. That week has been used in parts of West Africa, East Africa, and the lower Congo at least since the late seventeenth century. Today, it prevails throughout a long continuous belt stretching from eastern Ghana, through Togo, Benin, and Nigeria, to Cameroon.
The West African week is essentially a market cycle, and its users do not always make a conceptual distinction between its temporal form and the actual economic activity it regulates. The Tiv word kasoa and the Erik word urua, for example, denote both "week" and "market". The conceptual affinity between the two is also quote from some of the Ewe and Yoruba names of the days of the week--"market day", "second day of the market", "market day is tomorrow", and so on.
He traces the origin of our particular seven-day week back to two reinforcing sources: one, the seven-day Jewish week, which obviously became increasingly important as it morphed into the seven day Christian and Muslim weeks. Once the holy day is tied to a seven day cycle, there's really no changing it. Christians moved their holy day to Sunday, Muslims moved their holy day to Friday. But once you're on a seven day religious cycle, it's hard to move to an economic cycle of any other length, because the religious day (often a day with limited work) will end up interfering with the economic cycle. The Hebrew word for week, shavu'a, comes from the word for seven, sheva, and the word Shabbat comes from the root sh-b-t which means to cease from work. This possibly has its origin a larger near Eastern tradition--while the Jews had tradition of a holy day every seventh day, the Assyrians had a tradition of an evil day every seven days. Days set apart at the same interval, but for different purposes. If this is the case, it would be one of many customs the Hebrews shared with Assyrians, Babylonians, and other Semitic Mesopotamian and Levantine civilizations. Zerubavel argues that there's no evidence for a Hebrew seven day week before the Babylonian captivity, this depends heavily on how you date different sections of the Torah and, while his book doesn't go into those details, I find many of the assumptions he made unconvincing and would push the origins of the Hebrew week back well before 586 BCE when the Babylonians carried of Jerusalem's elite to exile.
The Assyrian, Ancient Persian, and Buddhist observance of first, eighth, fifteenth, and twenty-third days of a lunar cycle are really "quasi-weeks" in that they're still tied to the lunar cycle. The Israelites managed to separate the religious week from the moon entirely, and thereby allows it to endlessly repeat its own rhythm until the end of time. Zerubavel writes, "The dissociation of the week from the lunar cycle is, therefore, the most significant breakthrough in the evolution of this cake from its somewhat imperfect and predecessor."
In reality, the Jewish-cum-Christian-cum-Muslim cycle probably was only half of the equation that got us to our present day week. The seven day cycle, with each day associated with one the seven classical "planets" visible to the naked eye, Zerubavel says, likely originates only in the 2nd Century BCE.
While the Chaldeans (Babylonians) were the first two identify the seven "wanderers" (planets) which give names to our present weeks, Zerubavel says there's no evidence they ever gave those as names to their market-cycle:
While this [Chaldean] planetary theory evolved around 500 BC, and the earliest Babylonian horoscope dates from 409 BC, not one instance of a particular day being designated as "the day of the moon" or "the day of Venus," for example, has been found in pre-Hellenistic horoscopes.
While its origins--just like those of the Jewish seven-day cycle--may well lie in Mesopotamia, the astrological seven-day week actually came into being only in the aftermath of Alexander the Great's Conquest of western Asia, and was essentially a Hellenistic invention. It probably evolved sometime during the second century BC at the very heart of the Hellenistic world, namely Alexandria. (This might explain why our main classical source regarding this cycle attribute its invention to the Egyptian). It was only in Alexandria that three distinct practices that had evolved quite independently of the Chaldean planetary theory -- an astronomical practice of arranging the seven planets in a certain invariable order, a mathematical practice of subdividing the daily cycle into twenty-four hours, and an astrological theory known as the doctrine of "chronocratores" -- were nevertheless integrated with it so as to produce the astrological seven-day week in its final form. This cycle is therefore the product of the successful Hellenistic fusion of astronomy, astrology, and mathematics, as well as of the great cultural heritage of Egypt, Babylonia, and Greece.
The sequence of the seven days of the astrological week is essentially based on the arrangement of the seven [classical] planets in the fixed, invariable order Saturn-Jupiter-Mars-Sun-Venus-Mercury-Moon, a distinctively Hellenistic arrangement that evolved only in the second century BC. This planetary sequence corresponds to the order of orbital periods of the planets--Saturn's orbital period is approximately 29.46 years, Jupiter's 11.86 years, Mars's 686.98 days, the sun's (given the assumption that it is the sun that revolves around the earth rather than the other way around) 36.26 days, Venus's 224.70 days, Mercury's 87.97 days, and the moon's 29.53 days. [...]
This still does not explain why "the day of the moon", for example, always follows "the day of the sun" rather than "the day of Mercury". If we examine how the sequence of the seven days of the astrological week might have been derived from the order in which the Hellenistic astronomers arranged the seven planets, we are immediately struck by one obvious overall pattern. In order to derive the sequence of Planetary days from the series Saturn-Jupiter-Mars-Sun-Venus-Mercury-Moon, one must proceed by "planetary leaps" that consist of skipping two planets each time. Thus, beginning with Saturn, we would proceed to the sun (skipping Jupiter and Mars), from there to the Moon, [etc. says yodats, but think: that's Satur[n]day, Sunday, Mo[o]nday].
He later goes on to explain that this skipping likely evolved from a practice of associating hours with planets before associating days with them (by this point, the Egyptian idea of dividing a day into 24 hours was already widespread). So if a day started with a Saturn hour, then you'd go through three full cycles of planets making 21, then 22nd hour would be Saturn again, the 23rd Jupiter, 24th Mars, and then the first hour of the next day would be the Sun. Hence, because assigning planets to hours (the horo of horoscope has the same root as hour) predates the assigning of planets to days, by following this hourly cycle in the traditional order, we get a daily cycle in the "skip two" order (he's got the fully 168 hours assigned to planets in a seven day cycle on page 18, check for yourself on Google Books if you're confused; it's Figure 2).
(continued below)