Estimating k from an Initial Amount and One Later Measurement

Why One Later Measurement Can Determine a Rate

A continuous exponential model describes a quantity whose rate of change is proportional to its current amount. In this model, the amount at time $t$ is written $P(t)=P_0e^{kt}$, where $P_0$ is the initial amount and $k$ is the constant relative rate.

If the initial amount and one later observation are known, the model has only one unknown parameter: $k$. The observation tells us how much the quantity changed over a known time interval. Dividing the later amount by the initial amount gives the multiplicative change; taking a natural logarithm converts that change into the rate constant for the continuous model.

This calculation is useful in growth and decay applications, but its interpretation depends on the model assumptions. The formula gives the value of $k$ that makes the exponential model match the specified observation. A single observation does not, by itself, establish that the quantity truly follows a constant-relative-rate model at all times.

Formula for Estimating $k$

Definition: Suppose an exponential model is $P(t)=P_0e^{kt}$, with $P_0>0$. If the amount measured after a positive elapsed time $T$ is $P(T)=P_T>0$, the rate constant determined by those data is $$ k=\frac{1}{T}\ln\left(\frac{P_T}{P_0}\right). $$ Here $\ln$ denotes the natural logarithm, the inverse of the exponential function $e^x$.

The elapsed time $T$ is measured from the time at which $P_0$ was recorded. The later measurement must refer to the same quantity and use the same units as the initial amount. The ratio $P_T/P_0$ is dimensionless, so $k$ has units of inverse time, such as $\mathrm{h}^{-1}$, $\mathrm{day}^{-1}$, or $\mathrm{min}^{-1}$.

The required conditions are important. In an amount-based application, both amounts are positive, and the elapsed time is positive. A positive initial amount multiplied by $e^{kt}$ remains positive for every real $k$ and every real $t$, so a zero or negative later observation cannot be matched by this model with $P_0>0$. If $T=0$, the observation is not later than the initial measurement, and the formula would divide by zero.

Why the Formula Works

Substitute the measured later amount into the model. The algebra below isolates $k$; taking a logarithm is valid because the amount ratio is positive. The natural logarithm and exponential function are inverses, so $\ln(e^x)=x$.

Write the model at the observation time.
The model must match the later measurement $P_T$ at elapsed time $T$.

Divide by the initial amount.
Since $P_0>0$, division is permitted and leaves the exponential change factor.

Take natural logarithms and solve.
Because $P_T/P_0>0$, the logarithm is defined, and the equation becomes linear in $k$.

P_T=P_0e^{kT} $$ $$ \frac{P_T}{P_0}=e^{kT} $$ $$ \ln\left(\frac{P_T}{P_0}\right)=\ln(e^{kT})=kT $$ $$ k=\frac{1}{T}\ln\left(\frac{P_T}{P_0}\right).

This is a complete derivation of the estimate: it is the unique real value of $k$ that makes $P_0e^{kT}=P_T$, because the steps solve the equation without introducing or discarding any possible values.

Interpreting the Estimate

The ratio $P_T/P_0$ determines whether the fitted model represents growth or decay over the interval. The natural logarithm is positive when its argument is greater than $1$, zero when its argument is $1$, and negative when its argument is between $0$ and $1$. Since $T>0$, division by $T$ does not change the sign.

P_T>P_0 \Longrightarrow k>0, \qquad P_T=P_0 \Longrightarrow k=0, \qquad 0<P_T<P_0 \Longrightarrow k<0.

A positive $k$ means the fitted exponential model increases, while a negative $k$ means it decreases. If the two amounts are equal, the fitted rate is $k=0$, so the model is constant. The numerical size of $k$ must always be read together with its time unit: a rate expressed per hour is not numerically interchangeable with a rate expressed per day.

Worked Examples

Worked Example: Estimating a Growth Rate

A quantity starts at $500$ units and measures $650$ units $4$ hours later. Estimate $k$ in the continuous exponential model, rounding the rate to six significant digits, and check the prediction.

Identify the initial amount, later amount, and elapsed time:

P_0=500\text{ units},\qquad P_T=650\text{ units},\qquad T=4\text{ h}.

Form the later-to-initial ratio, then apply the formula:

\frac{P_T}{P_0}=\frac{650}{500}=1.3 $$ $$ k=\frac{1}{4}\ln(1.3) =\frac{0.262364264\ldots}{4} =0.065591066\ldots\ \mathrm{h}^{-1}.

To six significant digits, $k\approx0.0655911\ \mathrm{h}^{-1}$. The positive sign is consistent with the observed increase from $500$ to $650$ units.

Check the prediction using the unrounded rate first:

P(4)=500e^{(0.065591066\ldots)(4)} =500e^{\ln(1.3)} =500(1.3) =650\text{ units}.

Thus the unrounded estimate reproduces the observation exactly. If the six-significant-digit rate is used instead, the prediction is $500e^{(0.0655911)(4)}\approx650.000088$ units, which rounds to $650.000$ units to three decimal places. The tiny difference is due to rounding $k$.

Worked Example: Estimating a Decay Rate

A sample contains $800$ milligrams of a substance initially and $500$ milligrams after $6$ days. Estimate the continuous rate constant to six significant digits and verify the fit.

Use consistent amount units and record the elapsed time:

P_0=800\text{ mg},\qquad P_T=500\text{ mg},\qquad T=6\text{ days}.

The ratio and its logarithm are:

\frac{P_T}{P_0}=\frac{500}{800}=0.625 $$ $$ \ln(0.625)=-0.470003629\ldots $$ $$ k=\frac{-0.470003629\ldots}{6} =-0.0783339382\ldots\ \mathrm{day}^{-1}.

Rounded to six significant digits, $k\approx-0.0783339\ \mathrm{day}^{-1}$. The negative rate agrees with the decrease in the measured amount.

Substitute the unrounded estimate into the model to check the observation:

P(6)=800e^{(-0.0783339382\ldots)(6)} =800e^{\ln(0.625)} =800(0.625) =500\text{ mg}.

Using the rounded estimate gives $800e^{(-0.0783339)(6)}\approx500.000115$ mg, or $500.000$ mg to three decimal places. The original observation is recovered, up to the small effect of rounding.

Worked Example: A Small Increase over a Short Interval

A measured amount rises from $1200$ units to $1500$ units over $3$ minutes. Find the continuous rate constant to six significant digits and check the predicted amount after $3$ minutes.

First calculate the ratio:

P_0=1200\text{ units},\qquad P_T=1500\text{ units},\qquad T=3\text{ min} $$ $$ \frac{P_T}{P_0}=\frac{1500}{1200}=1.25.

Now take the logarithm and divide by the elapsed time:

\ln(1.25)=0.223143551\ldots $$ $$ k=\frac{0.223143551\ldots}{3} =0.0743811838\ldots\ \mathrm{min}^{-1}.

To six significant digits, the estimate is $k\approx0.0743812\ \mathrm{min}^{-1}$. It is positive because the later amount exceeds the initial amount.

Check by predicting the amount at $3$ minutes with the unrounded estimate:

P(3)=1200e^{(0.0743811838\ldots)(3)} =1200e^{\ln(1.25)} =1200(1.25) =1500\text{ units}.

Using the rounded rate gives $1200e^{(0.0743812)(3)}\approx1500.000073$ units, or $1500.000$ units to three decimal places. This confirms that the rounded estimate predicts the observed amount to the stated display precision.

A Common Mistake: Reversing the Ratio

The ratio must be later amount divided by initial amount. Reversing it changes the sign of the logarithm and therefore changes growth into decay, or decay into growth. For example, for an increase from $500$ to $650$ over $4$ hours, using $500/650$ gives the incorrect calculation:

k_{\text{wrong}} =\frac{1}{4}\ln\left(\frac{500}{650}\right) =\frac{1}{4}\ln\left(\frac{1}{1.3}\right) =-0.065591066\ldots\ \mathrm{h}^{-1}.

This negative value contradicts the observed increase. It also fails the required check: $500e^{k_{\text{wrong}}4}=500/1.3$, which is about $384.615$, not $650$. The correction is to use the later-to-initial ratio:

k=\frac{1}{4}\ln\left(\frac{650}{500}\right) =0.065591066\ldots\ \mathrm{h}^{-1}, \qquad 500e^{4k}=650.

A reliable habit is to check both the sign and the predicted amount. An increase should produce a positive estimate, a decrease should produce a negative estimate, and substituting the estimate into the model at time $T$ should recover the later measurement.

What One Observation Does and Does Not Establish

When the initial amount is known, one later measurement determines a single value of $k$ that fits that observation exactly, apart from numerical rounding. This makes the calculation straightforward, but it does not tell us whether the constant-relative-rate assumption is appropriate throughout the interval. The estimate is conditional on using the exponential model.

In practice, measurements may themselves be rounded or uncertain. The computed rate then reflects the recorded values, not necessarily an exact underlying rate. Keep the units consistent, retain unrounded values during intermediate steps, and round only the reported estimate. The prediction check is an algebraic consistency check: it confirms that the calculated rate solves the model equation for the supplied data.

In summary, form the ratio of the later amount to the initial amount, take its natural logarithm, and divide by the elapsed time. The result has inverse-time units, and its sign indicates growth or decay. Substituting the rate back into the exponential model provides a direct check against the observation.

Check Your Understanding

Use the model $P(t)=P_0e^{kt}$ and the formula derived above to answer these questions.

A quantity increases from $200$ units to $250$ units in $5$ hours. Write the expression for its estimated rate constant, including units.
If $P_T/P_0=0.4$ and $T>0$, what is the sign of $k$, and what does that sign indicate?
Why must the ratio in the logarithm be $P_T/P_0$, rather than $P_0/P_T$?
What problem arises if $P_T=0$ while $P_0>0$ in this exponential model?
Describe how substituting the estimated $k$ into $P(T)=P_0e^{kT}$ checks the calculation.

Estimating k from an Initial Amount and One Later Measurement

What You'll Learn

Why One Later Measurement Can Determine a Rate

Formula for Estimating \(k\)

Why the Formula Works

Interpreting the Estimate