Differential Equations: The Mathematics That Learned to Predict the Future
Every falling object, cooling cup of coffee, decaying isotope, and spreading epidemic obeys the same underlying logic: change does not happen all at once — it happens at a rate, and that rate is itself a function of the system’s current state. A differential equation is the formal sentence that records this fact. What follows traces how that sentence was invented out of necessity in the late seventeenth century, defines its working vocabulary in plain language, solves one completely by hand — the decay of radioactive carbon — and surveys the circuits, organisms, and economies still running on it today.
I. The Problem That Demanded a New Kind of Equation
A differential equation does not describe what a quantity is — it describes how fast that quantity is changing (its derivative). From that one piece of information, the entire future and past of the system can, in principle, be reconstructed.
By the middle of the 1600s, mathematicians could describe shapes — circles, parabolas, spirals — with remarkable algebraic precision. What they could not yet describe, in any formal language, was instantaneous change: the exact speed of a falling stone at one single moment, or the exact rate a heated rod cools at one single instant. Algebra was built for static relationships. Nature is rarely static.
Isaac Newton broke that impasse first, in private. Working around 1671 on a manuscript he called The Method of Fluxions and Infinite Series — not published until 1736, more than sixty years later — Newton classified what he called “fluxional equations” into three distinct classes, producing what historians now recognize as the first systematic treatment of differential equations.[1]
Gottfried Wilhelm Leibniz reached the same destination by an independent road, and arrived with the better map. On 11 November 1675, Leibniz wrote down the solution to what we would now call a differential equation — a date many historians treat as the field’s true birthday.[1] Leibniz is also the one who named the subject, coining the Latin aequatio differentialis for an equation relating the differentials dx and dy of two changing quantities.
Sixty-eight years of priority dispute aside, 1693 is the year both men published working solution methods for differential equations — the moment most historical accounts mark as the field’s formal inception as a distinct branch of mathematics, rather than a private notational trick.[1]
The relationship between the two men did not stay collegial. Newton’s allies at the Royal Society accused Leibniz of quietly absorbing unpublished material during a visit to London; Leibniz’s allies on the continent accused Newton’s camp of rewriting the historical record after the fact. Modern historians, with both men’s notebooks now fully catalogued, generally treat the pair as independent co-discoverers who reached overlapping ideas through genuinely different methods — Newton through physical reasoning about velocities and accelerations, Leibniz through a more abstract, symbolic notation that, not coincidentally, is the same notation still used in every differential equation in this article. The bitterness of the dispute slowed communication between English and continental mathematicians for nearly a generation, even as both traditions kept advancing the same subject in parallel.
Calculus alone did not force differential equations into existence, though — specific, stubborn physical puzzles did. Christiaan Huygens spent over a decade from 1659 onward chasing the isochrone: a curve along which a swinging pendulum keeps perfect time no matter how wide its swing. Solving it led directly to the first genuinely accurate pendulum clocks.[4]
In 1690, Jakob Bernoulli challenged the mathematical world to find the exact curve traced by a hanging chain — the catenary. Within a year, Johann Bernoulli, Huygens, and Leibniz had each solved it independently, each by setting up and solving a differential equation.[4] Johann Bernoulli escalated the family rivalry in 1696 with the brachistochrone problem — find the shape of a frictionless wire that lets a bead slide between two points in the least possible time — a challenge answered almost as fast as it was posed, by Newton, Leibniz, Jakob Bernoulli, and l’Hôpital.[4]
By 1714, Brook Taylor had derived the fundamental frequency of a stretched vibrating string by solving an ordinary differential equation relating tension, mass, and pitch — the mathematical ancestor of every tuned instrument since.[4] A generation later, Leonhard Euler — mentored by Johann Bernoulli from the age of fourteen — systematized the entire field. His two-volume 1736 Mechanica recast Newton’s laws of motion as differential equations, and he introduced integrating factors and the beta and gamma functions, building a toolkit still taught largely unchanged today.[2]
“Read Euler, read Euler, he is the master of us all.” Words attributed to Pierre-Simon Laplace, preserved only in a secondhand 1846 account — yet never disputed by the mathematicians who repeated it.
Euler’s productivity reshaped the subject’s center of gravity. Across more than 800 papers and books, he converted differential equations from a scattered set of clever tricks for individual curves into a general theory with reusable methods — among them the separation-of-variables technique Section III will use to date a fossil. By the time Euler died in 1783, every quantity that changed smoothly over time — a planet’s position, a vibrating string’s height, a cooling kettle’s temperature — had, in principle, a differential equation waiting to describe it; the eighteenth and nineteenth centuries would spend themselves finding which equation fit which phenomenon.
That search for “which equation fits which phenomenon” is this article’s entire second half. Section II first fixes the vocabulary needed to read any of these equations; Section III then runs the search to its conclusion for one specific, falsifiable case.
Differential equations were forged to solve real problems — pendulum clocks, hanging chains, racing beads, vibrating strings — not invented in the abstract.
II. The Working Vocabulary
Before the decay problem in Section III can mean anything, ten terms need fixed, working definitions — not textbook abstractions, but the specific sense in which this article uses them.
These distinctions are not academic hair-splitting. Historically, the order and linearity of a differential equation determined how much specialized machinery a mathematician needed before electronic computers existed: a first-order separable equation like the one in Section III could be solved by hand in an afternoon, while higher-order nonlinear systems often resisted exact solution for centuries and eventually pushed Henri Poincaré, in the 1880s, toward inventing entirely new qualitative methods rather than continuing to search for a formula at all.
- DerivativeThe instantaneous rate of change of a quantity with respect to another, almost always time. Written dy/dt or y′.
- Differential equationAny equation that relates a quantity to its own derivative or derivatives.
- OrderThe highest derivative appearing in the equation. dN/dt = −kN is first-order because only the first derivative appears.
- Ordinary differential equation (ODE)A differential equation involving derivatives with respect to only one variable, usually time.
- Partial differential equation (PDE)A differential equation involving derivatives with respect to two or more variables — for instance, position and time — such as the equations governing heat flow or quantum wavefunctions.
- Linear vs. nonlinearAn equation is linear if the unknown function and its derivatives appear only to the first power and are never multiplied together. Nonlinear equations, common in weather and population models, can produce chaotic behavior.
- General solutionA family of functions, containing one or more arbitrary constants, that satisfies the differential equation.
- Initial conditionA known value of the quantity at a starting time, used to select one specific (“particular”) solution out of the general family.
- Separable equationA first-order equation in which the variables can be algebraically sorted onto opposite sides before integrating — the exact technique used in Section III.
- Equilibrium solutionA constant solution at which the rate of change is permanently zero — a steady state the system can settle into.
Every one of the ten terms above will reappear, by name, in the worked derivation that follows. Watch in particular for order, separable, and initial condition — between them, they explain every move made in Section III.
Order tells you how much memory an equation needs — just the current state, or the current state and its rate of change. Linear vs. nonlinear tells you whether the system can ever behave unpredictably.
Ten words — derivative, order, linear, separable, and equilibrium among them — supply the entire grammar needed to read any differential equation.
III. Dating a Bone with One Differential Equation
Solving by separation of variables (Definition 9) is careful algebra plus integration: isolate the changing quantity, integrate both sides, then use one known data point — the initial condition — to pin down the one remaining constant.
Radiocarbon dating rests on a single fact, confirmed across thousands of laboratory measurements: a radioactive isotope decays at a rate proportional to how much of it is currently present. Twice as much carbon-14 means twice as many atoms decaying per second; half as much means half the decay rate.
Notice, before the steps begin, that none of them requires calculus beyond a first course. Separating variables is ordinary algebra; the integral of 1/N is a standard result memorized in any introductory class; and exponentiating to undo a logarithm is the same move used to solve for x in an equation like 10x = 1000. The only genuinely new idea is step five — trading the abstract constant k for a measurable half-life — and that single substitution is what turns a pure mathematics exercise into a laboratory technique.
- Write the rate law as an equation. Let N(t) be the number of carbon-14 atoms remaining at time t. “The rate of change of N is proportional to N” becomes a differential equation directly: dN/dt = −kN where k is a positive constant (the decay constant) and the minus sign records that N is shrinking.
- Separate the variables. Every N belongs on one side; divide by N and multiply by dt: dN / N = −k dt
- Integrate both sides. Integrating the left side with respect to N and the right with respect to t: ln|N| = −kt + C
- Exponentiate and apply the initial condition. Raising e to both sides clears the logarithm. At t = 0, N must equal the starting amount N₀, which pins down the constant: N(t) = N₀ e−kt
- Connect k to something measurable: the half-life. k itself can’t be read off a lab instrument, but the half-life T (the time for half a sample to decay) can. Setting N(T) = N₀/2 and solving gives: k = ln(2) / T
- Plug in carbon-14’s measured half-life. Carbon-14’s half-life has been measured at 5,730 years, so: k = ln(2) / 5730 ≈ 0.000121 per year
- Use the equation to read a fossil’s age. If a bone sample contains 25% of the carbon-14 found in living tissue, then N(t)/N₀ = 0.25 = e−0.000121t. Taking the natural log of both sides and solving for t: t = ln(4) / 0.000121 ≈ 11,460 years — almost exactly two half-lives, exactly as expected, since 25% is one half of one half.
It is worth checking that arithmetic against intuition before trusting it. Two half-lives should leave exactly one quarter of the original carbon-14 — 50% of 50% — and the derivation’s own numbers confirm it: at t = 11,460 years, e−0.000121×11460 works out to almost exactly 0.25. A differential equation that fails this kind of sanity check is either mis-derived or mis-applied; one that passes earns the right to be trusted on dates no living person personally observed.
That seven-step path — write the rate law, separate, integrate, exponentiate, calibrate against one measured fact, solve for the unknown — is identical whether the decaying quantity is a radioisotope, a drug concentration in a bloodstream, or charge draining from a capacitor.
N(t) = N₀e−kt — hover or tap the curve to enlarge.
Every radiocarbon date traces back to one separable differential equation and one measured half-life: 5,730 years.
IV. From Ink to Infrastructure: Where the Equation Lives Now
The same first-order rate law solved by hand in Section III reappears, with only the constant and the sign changed, across engineering, medicine, and physics.
Newton’s Second Law. F = ma is itself a second-order differential equation in disguise — acceleration is the second derivative of position. Every flight simulator, crash-test model, and orbital-trajectory calculation is, underneath, a differential equation being solved numerically thousands of times per second; the Apollo 11 descent to the lunar surface, for instance, was guided by onboard solutions of exactly this kind of equation, recomputed continuously as new tracking data arrived.
RC Circuits. A charging capacitor obeys dV/dt = (Vs − V)/RC — the same exponential shape as decay, just running toward a target voltage instead of toward zero. It is the equation that times every camera flash and pacemaker pulse, and an engineer can speed it up or slow it down simply by swapping in a different resistor.
Newton’s Law of Cooling. A cooling body obeys dT/dt = −k(T − Tambient), used by forensic investigators estimating time of death from body temperature, and by every oven thermostat deciding when to fire the heating element again. Measuring a body’s temperature twice, an hour apart, against a known room temperature, supplies enough data to solve for both the time of death and the constant k at once.
Logistic Growth & Epidemiology. Population and epidemic models replace the constant k with a term that depends on how crowded or how infected the population already is, producing the S-shaped logistic curve behind bacterial culture growth and the SIR models public-health agencies used to track outbreak reproduction rates. The same S-curve shape describes a yeast colony filling a petri dish and a new technology reaching market saturation — the equation does not distinguish between the two.
Quantum Mechanics. At the smallest scales, the Schrödinger equation — a partial differential equation rather than an ordinary one — plays the same role for a particle’s wavefunction that dN/dt = −kN plays for a decaying nucleus: both are statements about how fast something changes, solved with the algebraic toolkit built between the 1670s and the 1730s. Unlike the carbon-14 equation, it tracks a probability wave rather than a fixed count of atoms, but the underlying question — how fast does this state change? — is unchanged.
Continuously Compounded Interest. A bank balance earning continuous interest obeys dP/dt = rP — the same equation, with the same sign, that governs an unchecked population explosion. It is the reason a savings account and an unmanaged bacterial culture share the same growth curve, and the reason compound interest is sometimes called, half-jokingly, the eighth wonder of the world.
Not every differential equation surrenders this easily. The instant a rate of change depends on the system in a nonlinear way — friction proportional to velocity squared, a population whose growth rate falls as resources run short, three gravitating bodies pulling on each other simultaneously — exact pencil-and-paper solutions usually vanish, and the same kind of equation that predicts a fossil’s age with confidence can instead produce the genuinely unpredictable behavior known as chaos. Henri Poincaré discovered this while studying the three-body problem in the 1880s, more than two centuries after Newton wrote down the first fluxional equations — a reminder that inventing the language of change and fully understanding what that language can say are two different achievements.
“It’s a mysterious and marvelous fact that our universe obeys laws of nature that always turn out to be expressible in the language of calculus as sentences called differential equations.” Steven H. Strogatz, Infinite Powers (2019)
Hover or tap any node group to enlarge the map.
None of these six applications required inventing new mathematics. Each one only needed someone willing to notice that a system’s rate of change depended on its own current state — the same noticing that drove Huygens to the pendulum clock and Bernoulli to the hanging chain three centuries earlier.
Changing only the sign and the constant in dQ/dt = ±kQ converts the same equation between a charging capacitor, cooling coffee, a clearing drug dose, and a decaying isotope (proportional rate law). The mathematics never asks what Q represents.
The one equation solved in Section III also governs capacitors, cooling coffee, and clinical drug dosing — only the sign and constant change.