Estimating k from Two Measurements at Different Times

Why Two Measurements Can Be Enough

A growth or decay process is often observed after it has already begun. You may know the amount at two later times but have no measurement at time zero. The continuous exponential model still allows you to estimate its rate constant: the unknown starting amount cancels when the two measurements are compared.

This is useful when the data come from observations made during a study rather than from a known initial state. The calculation depends on an assumption: over the interval being studied, the process follows a constant-relative-rate model.

Key idea: If $P(t)=P_0e^{kt}$, then the ratio of two measurements depends on the rate constant and the time interval, but not on $P_0$.

The Rate Formula from Two Measurements

Suppose $P(t)$ is a positive quantity described by the continuous exponential model $P(t)=P_0e^{kt}$, where $P_0$ is a positive reference amount and $k$ is the continuous rate constant. Let $t_1$ and $t_2$ be distinct times, and suppose the measured amounts are $P(t_1)$ and $P(t_2)$.

Dividing the amount at the second time by the amount at the first gives $P(t_2)/P(t_1)=e^{k(t_2-t_1)}$. Because both amounts are positive, the ratio is positive and its natural logarithm is defined. Taking the natural logarithm and dividing by the nonzero time difference gives the estimate.

Formula: For two positive measurements at distinct times, under the constant-relative-rate model, $$ k=\frac{\ln\!\left(P(t_2)/P(t_1)\right)}{t_2-t_1}. $$ If the measurements are observed data rather than exact model values, this calculation gives an estimate of $k$.

The times should be paired with their corresponding measurements, and their order must be kept consistent. A convenient convention is $t_2>t_1$, so the time difference is positive. Under that convention, a larger later amount gives a positive $k$, while a smaller later amount gives a negative $k$. The units of $k$ are the reciprocal of the time units: if time is measured in weeks, $k$ is measured in inverse weeks. The measurement units cancel in the ratio.

Why the Formula Works

The derivation uses the exponential solution to the differential equation $dP/dt=kP$. The essential point is cancellation: the same reference amount appears in both model values.

Write the model at each time.
$P(t_1)=P_0e^{kt_1}$ and $P(t_2)=P_0e^{kt_2}$.

Divide the later value by the earlier value.
Since $P_0>0$, it cancels, leaving $P(t_2)/P(t_1)=e^{k(t_2-t_1)}$.

Isolate the rate constant.
Taking natural logarithms gives $\ln(P(t_2)/P(t_1))=k(t_2-t_1)$. Since the times are distinct, divide by $t_2-t_1$.

This is a complete derivation when the two values follow the model exactly. For measured data, the same steps produce a rate estimate: they find the constant rate that makes the model reproduce the observed ratio over that interval. The calculation does not establish that the process truly had a constant relative rate; that is a modeling assumption.

Worked Examples

Worked Example: Estimating a Growth Rate

A population is measured at 30 units at $t=2$ weeks and 120 units at $t=6$ weeks. Estimate its continuous rate constant, assuming exponential growth over this interval.

Here $P(t_1)=30$, $t_1=2$, $P(t_2)=120$, and $t_2=6$. The elapsed time is $6-2=4$ weeks, and the measurement ratio is $120/30=4$. Therefore,

k=\frac{\ln(120/30)}{6-2} =\frac{\ln 4}{4} \approx \frac{1.3863}{4} \approx 0.3466\ \text{week}^{-1}.

The estimate is positive, consistent with the population increasing. Check the result by recovering the observed ratio:

e^{k(6-2)} =e^{(\ln 4/4)(4)} =e^{\ln 4} =4 =\frac{120}{30}.

Thus the estimated rate reproduces the measured fourfold change over four weeks.

Worked Example: Estimating a Decay Rate

A substance has an amount of 240 milligrams at $t=1$ hour and 60 milligrams at $t=5$ hours. Estimate the continuous rate constant.

The elapsed time is $5-1=4$ hours. The later-to-earlier ratio is $60/240=1/4$. Substitution gives

k=\frac{\ln(60/240)}{5-1} =\frac{\ln(1/4)}{4} =-\frac{\ln 4}{4} \approx -\frac{1.3863}{4} \approx -0.3466\ \text{hour}^{-1}.

The rate is negative, as expected when the amount falls with time. To check, use the estimated rate to calculate the model's ratio across the four-hour interval:

e^{k(5-1)} =e^{(-\ln 4/4)(4)} =e^{-\ln 4} =\frac{1}{4} =\frac{60}{240}.

The ratio agrees with the measurements. The rate constant is expressed per hour because the time values were given in hours.

Worked Example: Estimating a Rate from a Modest Change

A quantity is measured as 75 units at $t=3$ days and 60 units at $t=9$ days. Estimate the continuous rate constant and check it against the data.

The interval is $9-3=6$ days. The later-to-earlier ratio is $60/75=4/5=0.8$. Hence

k=\frac{\ln(60/75)}{9-3} =\frac{\ln(0.8)}{6} \approx \frac{-0.2231}{6} \approx -0.0372\ \text{day}^{-1}.

The negative sign indicates decay, although the decrease is smaller than in the previous example. Checking without rounding the estimate,

e^{k(9-3)} =e^{(\ln(0.8)/6)(6)} =e^{\ln(0.8)} =0.8 =\frac{60}{75}.

The unrounded formula recovers the exact observed ratio. The displayed decimal is rounded to four decimal places, so substituting that rounded value will give a ratio close to, rather than exactly, 0.8.

Conditions and Interpretation

Before using the formula, check that the assumptions and data fit the calculation:

The times are distinct. If $t_2=t_1$, the denominator is zero, so this formula cannot estimate a rate from those observations.
The measured amounts are positive. The exponential model with a positive reference amount gives positive values, and the logarithm requires a positive ratio.
The measurements refer to the same quantity. Their units must agree so that the ratio is meaningful and unitless.
The model is intended to apply between the measurements. The result is an estimate under the assumption of a constant relative rate, not proof that the assumption is appropriate.

When $t_2>t_1$, the sign also provides a quick interpretation. If $P(t_2)>P(t_1)$, then $P(t_2)/P(t_1)>1$, so its logarithm and $k$ are positive. If $0<P(t_2)<P(t_1)$, the ratio is between zero and one, its logarithm is negative, and $k<0$. Equal positive amounts give a ratio of one and $k=0$.

The estimated $k$ is an average constant relative rate over the chosen interval under the model. Changing the interval can change the estimate if the real process does not maintain a constant relative rate. Assessing whether the data support that assumption requires more than calculating a single two-point estimate.

A Common Mistake: Reversing Only Part of the Calculation

Suppose the amount decreases from 80 units at $t=2$ to 20 units at $t=6$. A common error is to use the later-to-earlier amount ratio but reverse the time difference, writing $\ln(20/80)/(2-6)$. This gives a positive value, which conflicts with the observed decrease.

\frac{\ln(20/80)}{2-6} = \frac{\ln(1/4)}{-4} = \frac{-\ln 4}{-4} = \frac{\ln 4}{4} >0.

The problem is that the amount ratio and time difference no longer use the same order. Pair the later amount with the later time in the numerator and denominator, or use the earlier amount and time consistently. With $t_2=6$ and $t_1=2$, the correct calculation is

k=\frac{\ln(20/80)}{6-2} =\frac{\ln(1/4)}{4} =-\frac{\ln 4}{4} \approx -0.3466.

Another error is to use the difference in amounts, such as $(20-80)/(6-2)$, as if it were $k$. That expression is an average change in amount per unit time, with units of amount per time. The continuous rate constant instead comes from the logarithm of a dimensionless ratio divided by elapsed time, so its units are inverse time.

Using the Estimate Carefully

A two-measurement estimate is especially straightforward when the data follow the model exactly. In practical measurements, values may be rounded or affected by measurement variation, so the calculated value should be described as an estimate. The arithmetic still has a useful built-in check: using the unrounded estimate in $e^{k(t_2-t_1)}$ must return the ratio $P(t_2)/P(t_1)$ used to compute it.

Keep the ratio and the time interval aligned, carry units through the calculation, and round only when reporting the final estimate. These habits make it easier to interpret the result and to catch sign errors before they affect later calculations.

Check Your Understanding

Use the two-measurement formula and state the units and sign of each estimate where appropriate.

A quantity is 10 units at $t=1$ hour and 40 units at $t=4$ hours. What is the estimated $k$?
A quantity is 90 milligrams at $t=2$ hours and 30 milligrams at $t=8$ hours. What is the estimated $k$?
Why does the unknown reference amount cancel when measurements at two times are divided?
What happens to the formula if the two measurement times are equal?
Explain why the expression $(P(t_2)-P(t_1))/(t_2-t_1)$ is not the continuous rate constant $k$.

Estimating k from Two Measurements at Different Times

What You'll Learn

Why Two Measurements Can Be Enough

The Rate Formula from Two Measurements

Why the Formula Works