The "dignity" of rational beings requires that they accept no law that they would not themselves have enacted. Commodities and products that serve physical needs and desires have "prices" in the market. By contrast, qualities that constitute people as ends in themselves have an intrinsic, absolute value--they have "dignity." The dignity of morality is the criterion for people to serve as lawmakers in the kingdom of ends.
Thus the principle of morality may be formulated in three distinct but interrelated ways: (1) in terms of the form of universality (act such that your maxim could become universal law); (2) in terms of their purpose or "end" (act such that all rational beings are respected as ends in themselves); and (3) in terms of a complete social system (act such that your maxim could be law in the kingdom of ends). An absolutely good will must never be in conflict with itself; its actions must have the intrinsic value of universal laws of reason. The purposes of an absolutely good will must never be relative only to certain ends, but must rather have the intrinsic value of ends that could be recognized by all rational beings. Consequently, the absolutely good will must choose its maxims as though it were a law-giver in the kingdom of ends--even though there is no guarantee that the contingencies of nature and the actions of other people will not prevent the establishment of such a kingdom.
When rational beings pursue morality and the kingdom of ends, they elevate themselves above the demands of nature and of their material circumstances. They thus establish the independence, or "autonomy," of their will. By contrast, when a person's goals are determined by something other than universal law, their will is "heteronomous"--it depends on external factors in determining its goals.
Other philosophical systems have made the mistake of advancing bases for morality that would in fact render the will heteronomous. "Empirical" principles--principles oriented toward some outcome in the physical world-- cannot be the basis of morality, because they are always heteronomous; even when the goal is personal happiness, concerns about particular outcomes or courses of events can never have the status of universal laws of nature. "Rational" principles like the will of God are likewise heteronomous because they do not come from pure concepts of reason; we have no notion of divine perfection other than that which we derive from our own moral concepts. Anytime someone does something in order to attain something else--whether that something else is happiness or perfection or the satisfaction of some physical need or desire--the person's will is determined by that something else; the will is heteronomous, and the maxim of the action makes sense only in particular circumstances, not as a universal law of nature.
We still have not shown that there is a law that compels us in practice to abide by the categorical imperative. We have shown, however, that our general presuppositions about morality (the idea that moral actions are undertaken for the sake of duty alone) are based on a notion of the autonomy of the will.
It may seem confusing that Kant precedes and follows his discussion of morality and the "will" with a disclaimer that he has not established that the categorical imperative has binding force for rational beings. Recall the provisional nature of Kant's argument in this book: it is only a "grounding" for the metaphysics of morals, not a full metaphysics of morals, let alone a complete analysis of "practical" (moral) reason and its role in our lives. Kant started off in Chapter 1 with the presupposition that people generally think of moral actions as actions performed for the sake of duty alone. He then developed an account of the "moral law" that may be based on this notion of duty and morality. In the first half of Chapter 2 he reformulated this moral law in terms of the categorical imperative. In the remainder of Chapter 2 he develops an account of the implications the moral law must have for the will of rational beings. Only in Chapter 3 will Kant explain that morality may be based on the concept of free will. As we will see, Kant qualifies even this statement by noting that the concept of free will cannot fully explain why we feel compelled to behave morally.