Tangents To A Circle From Nearby Secants

From Chords to a Tangent

A circle is a useful setting for seeing how a tangent direction emerges from nearby secants. A secant line passes through two points of the circle; as one point moves toward a fixed point, the secant line may approach a definite limiting direction. That direction is the tangent direction.

For a circle, the limit also reveals a geometric relationship: the radius from the center to the point of tangency is perpendicular to the tangent. We will derive this relationship from the circle’s equation, rather than rely on a diagram alone.

The Circle and Its Secants

Let a circle have center \((h,k)\) and radius \(r>0\). Its equation is \((x-h)^2+(y-k)^2=r^2\). Choose a point \(P=(a,b)\) on the circle, so \((a-h)^2+(b-k)^2=r^2\). A second, distinct point \(Q=(x,y)\) on the circle determines a secant line through \(P\) and \(Q\).

It is convenient to measure coordinates relative to the circle’s center. Write \(u=a-h\) and \(v=b-k\). The radius from the center to \(P\) has direction vector \((u,v)\), and \(u^2+v^2=r^2\). For a nearby point \(Q\), write \(x=a+\Delta x\) and \(y=b+\Delta y\). Then \(Q\) approaches \(P\) when both increments approach zero.

Deriving the Secant-Slope Limit

Because \(P\) and \(Q\) lie on the same circle, their center-relative coordinates have equal squared lengths. Subtracting the equation for \(P\) from the equation for \(Q\) gives an exact relation between the increments:

Suppose first that \(v\ne0\), meaning that \(P\) is not at the leftmost or rightmost point of the circle. Nearby distinct points on the circle have \(\Delta x\ne0\): the vertical line \(x=a\) meets the circle only at \(P\) and its reflection across the horizontal line through the center, and that reflected point is a fixed positive distance from \(P\). Thus, for points sufficiently close to \(P\), the secant slope is defined.

Since \(\Delta y=m_{\text{sec}}\Delta x\), divide the increment relation by \(\Delta x\) and solve for \(m_{\text{sec}}\). For nearby points, \(2v+\Delta y\ne0\), since it approaches the nonzero number \(2v\).

As \(Q\) approaches \(P\), the increments approach zero. The denominator approaches \(2v\), which is not zero, so the secant slopes approach a finite limit:

Worked Examples: Computing the Limit

When the Tangent Is Vertical

The finite-slope formula requires \(b-k\ne0\). The remaining points have \(b=k\), so \(P\) lies at the leftmost or rightmost point of the circle. In this case \(u=a-h\ne0\), because \(u^2+v^2=r^2\), \(v=0\), and \(r>0\).

With \(v=0\), the increment relation becomes \(\Delta x(2u+\Delta x)=-(\Delta y)^2\). For nearby points distinct from \(P\), \(\Delta y\ne0\); otherwise the circle equation would give only \(P\) or the opposite endpoint, which is not arbitrarily close to \(P\). Since \(2u+\Delta x\) approaches the nonzero number \(2u\), divide by \(\Delta y(2u+\Delta x)\):

Thus the horizontal component of the secant direction becomes negligible compared with its vertical component. The secant directions approach vertical, so the tangent is vertical. Its line is \(x=a\). The radius is horizontal at these endpoints, so its direction \((u,0)\) is perpendicular to the vertical tangent direction \((0,1)\).

Why the Radius Is Perpendicular to the Tangent

The limit formula gives a direct proof when \(v\ne0\). The radius direction is \((u,v)\), and a direction vector for a line with slope \(m_{\text{tan}}\) is \((1,m_{\text{tan}})\). Their dot product is:

Two nonzero vectors with dot product zero are perpendicular, so the radius is perpendicular to the tangent. This also covers \(u=0\): the formula gives tangent slope \(0\), and the radius is vertical while the tangent is horizontal. When \(v=0\), the separate limiting argument above gives a vertical tangent and a horizontal radius, again perpendicular.

It is sometimes useful to express the tangent line without dividing by \(v\). The line through \(P\) whose direction is perpendicular to the radius has equation:

When \(v\ne0\), rearranging gives \(y-b=-\frac{u}{v}(x-a)\), the finite-slope tangent line. When \(v=0\), \(u\ne0\) and the equation reduces to \(x=a\), the vertical tangent. This form handles both cases without assigning a finite slope to a vertical line.

A Common Pitfall: The Negative Reciprocal Has a Condition

For two nonvertical perpendicular lines, their slopes are negative reciprocals. That description cannot be applied when one of the slopes is undefined. For example, on the unit circle at \(P=(0,1)\), the radius slope would be calculated as \((1-0)/(0-0)=1/0\), which is undefined. Calling the tangent slope its negative reciprocal is therefore not a valid calculation.

What the Limit Reveals

The secant calculation does more than provide a formula for one circle. It shows why tangent direction is local: it is determined by the way nearby points on the curve sit relative to the point of contact. For a circle, subtracting the equations of two points leads to a simple relation whose limiting direction is perpendicular to the radius.

The exceptional cases are part of the result, not exceptions to the geometry. At the top and bottom of a circle, the tangent is horizontal and the radius is vertical. At the leftmost and rightmost points, the tangent is vertical and the radius is horizontal. Distinguishing a finite slope from a vertical direction keeps the limit argument correct at every point of the circle.