No trigonometric function is one-one on its entire natural domain, since periodicity means infinitely many angles share the same function value — so none can be inverted over all of R (or R minus its excluded points). Examining each function's graph, however, shows a specific restricted interval on which it becomes one-one and onto, and it is only on that restricted interval that an inverse is defined: sine on [−2π,2π] gives sin−1:[−1,1]→[−2π,2π]; cosine on [0,π] gives cos−1:[−1,1]→[0,π]; tangent on (−2π,2π) gives tan−1:R→(−2π,2π); and similarly restricted domains give cosec−1, sec−1, and cot−1. For x in the domain of an inverse function, the unique angle θ in its restricted range with (for example) sinθ=x is called the principal value of sin−1x — even though other angles outside the range also have sine x, only the one inside the agreed restricted interval counts as the principal value, which is why sin−1(siny)=y only holds when y already lies in [−2π,2π], and must otherwise first be reduced to an equivalent angle that does.