Menu
Sign up Sign in

AI HL · 12.6

Exponents & Logarithms

12.6.4

Logarithmic scales on graphs

12.6.4

Logarithmic scales on graphs

Target: I can explain why data over a very wide range is plotted on a log scale, and write the log10 of its values. (+4 more)

By the end of this section

Key points

Why log scales

  • Data that runs from very small to very large (areas, populations, masses) bunches up on a normal axis.
  • A log scale plots \(\log_{10}\) of the value. Multiplying a value by 10 adds 1 to its log, so equal steps mean equal factors.
  • A scale in logs of y only is semi-log. A scale in logs of both x and y is log-log.

Exponential: semi-log graph

  • \(y=a\times b^{x}\) gives \(\log y=\log a+x\log b\). This is a straight line: log y against x.
  • The gradient is \(\log b\) and the y-intercept is \(\log a\). So \(b=10^{\text{gradient}}\) and \(a=10^{\text{intercept}}\).
  • The intercept is at x = 0. If b < 1 the line slopes down.

Power: log-log graph

  • \(y=a\,x^{k}\) gives \(\log y=\log a+k\log x\). This is a straight line: log y against log x.
  • The gradient is k and the y-intercept is \(\log a\), so \(a=10^{\text{intercept}}\).
  • The intercept is where log x = 0, that is x = 1.

Choosing and using the model

  • Plot both graphs, or find r for both. The straighter one (r closer to 1 or −1) shows the relationship.
  • Use the calculator's linear regression on the logged values. Do not round the gradient and intercept before you undo the logs.
  • Predict with the model, not with the log graph. Say that predictions outside the data are less reliable.

On the calculator

Logging a column

  • Put the data in two columns of a spreadsheet or list.
  • Make a new column with the formula log(x) or log(y).
  • Name the new columns so you can use them in the regression.

Screenshot and video to come

Linear regression on logged data

  • Choose linear regression, y = mx + b, with the logged columns as x and y.
  • Read the gradient m, the intercept b and the correlation coefficient r.
  • Compare r for log y against x and log y against log x.

Screenshot and video to come

Back from logs to the model

  • Calculate 10 to the power of the gradient (semi-log) or intercept.
  • Keep the full values on the calculator and round only the final a, b or k.

Screenshot and video to come

Explore: which axes make it straight?

Choose a set of data, then change the axes. One choice of axes puts the points on a straight line. Predict which before you press it.



Plotted
Gradient of the line of best fit
Intercept
Correlation, r

Semi-log means log y against x. Log-log means log y against log x. The logs are to base 10.

Choose the axes.

  1. On linear axes, where do the points bunch up? Why does a log scale spread them out?
  2. For the bacteria, which axes give a straight line? For the animal energy data? What kind of model does each straight line mean?
  3. Compare r on the three kinds of axes. Does a higher r always mean the right model? What else do you look at?
  4. On the straight graph, the gradient is not b itself. How do you get b (or a) from the gradient and the intercept? Check your answer against the model shown.

Worked examples

Problems like the ones in class. Pick one, try it, then open the working.
Why a wide range needs a log scale

The body masses of six animals are 0.02 kg, 0.5 kg, 5 kg, 70 kg, 500 kg and 4000 kg. (a) Write the log10 of each mass. (b) Say why the logs are easier to plot than the masses.

Working
  1. Take log10 of each

    From the calculator: −1.70, −0.301, 0.699, 1.85, 2.70, 3.60

    Use the log key. Round to 3 significant figures.

  2. Mass (kg)0.020.55705004000log10 of mass−1.70−0.3010.6991.852.703.60
    The table

    Masses on the top row, their logs on the bottom row.

  3. Why it helps

    Masses run from 0.02 to 4000. Logs run from −1.70 to 3.60.

    On an axis from 0 to 4000 the small masses all sit on top of 0. The logs are spread over about 5 units, and each step of 1 means a factor of 10.

Video walkthrough coming
A straight line on a semi-log graph: exponential

The number of bacteria N (thousands) in a culture is measured every hour. (a) Plot log10 N against t and fit a straight line with the calculator. (b) Write the model in the form \(N=a\times b^{t}\).

t (hours)012345N (thousands)3.65.439.321423.436.3
Working
  1. 1234560.511.5tlog10 N
    Log the y values

    log10 N: 0.556, 0.735, 0.969, 1.15, 1.37, 1.56

    Keep t as it is. Only y is logged: a semi-log graph.

  2. Linear regression

    From the calculator: gradient \(m=0.203\), intercept \(c=0.549\), \(r=0.9995\)

    r is close to 1, so the points are close to a straight line: the data is exponential.

  3. Undo the logs

    \(b=10^{m}=10^{0.203}=1.60\), \(a=10^{c}=10^{0.549}=3.54\)

    \(\log y=\log a+x\log b\): the gradient is \(\log b\) and the intercept is \(\log a\).

  4. The model

    \(N=3.54\times1.60^{t}\)

    Thousands of bacteria after t hours.

Video walkthrough coming
A straight line on a log-log graph: power

The energy use R (watts) of five animals of body mass m (kg) is recorded. (a) Plot log10 R against log10 m and fit a straight line with the calculator. (b) Write the model in the form \(R=a\,m^{k}\).

m (kg)0.521050300R (watts)6.0616.357.9184728
Working
  1. −0.50.511.522.50.511.522.53log10 mlog10 R
    Log both columns

    log10 m: −0.301, 0.301, 1.00, 1.70, 2.48 log10 R: 0.782, 1.21, 1.76, 2.26, 2.86

    Both x and y are logged: a log-log graph.

  2. Linear regression

    From the calculator: gradient \(=0.750\), intercept \(=1.00\), \(r=0.9999\)

    r is close to 1: the data follows a power law.

  3. Read k and a

    \(k=0.750\), \(a=10^{1.00}=10.0\)

    \(\log y=\log a+k\log x\): the gradient is k and the intercept is \(\log a\).

  4. The model

    \(R=10.0\,m^{0.750}\)

    R in watts, m in kilograms.

Video walkthrough coming
Which graph is straight?

Data are recorded for x and y. (a) Decide whether log y against x or log y against log x is a straight line. (b) Write the model, and find the exponent.

x124816y31248192768
Working
  1. Log the values

    log10 x: 0, 0.301, 0.602, 0.903, 1.20 log10 y: 0.477, 1.08, 1.68, 2.28, 2.89

    Try both graphs. Doubling x always adds 0.301 to log x.

  2. Test both

    log y against log x: \(r=1.000\). log y against x: \(r=0.9333\).

    A perfect straight line has r = 1. The log-log graph is the straight one, so the relationship is a power.

  3. Gradient and intercept

    Gradient \(=2.00\), intercept \(=0.477\), so \(k=2\) and \(a=10^{0.477}=3\)

    Check: log 3 = 0.477.

  4. Model

    \(y=3x^{2}\)

Video walkthrough coming
Reading a model from a semi-log graph

The graph shows \(\log_{10}y\) against x for a model \(y=a\times b^{x}\). The line passes through (0, 1.2) and (4, 2.2). (a) Find a and b. (b) Find the value of x when \(y=1000\).

123456780.511.522.533.54xlog10 y(0, 1.2)(4, 2.2)
Working
  1. Gradient

    \(m=\dfrac{2.2-1.2}{4-0}=0.25\)

    The gradient of a semi-log line is \(\log b\).

  2. Intercept

    \(c=1.2\)

    The line meets the vertical axis at \(\log a\).

  3. Undo the logs

    \(b=10^{0.25}=1.78\), \(a=10^{1.2}=15.8\)

    The model is \(y=15.8\times1.78^{x}\).

  4. Predict x

    \(\log_{10}1000=3=1.2+0.25x\), so \(x=7.20\)

    Work in log form: it is a linear equation.

Video walkthrough coming
Reading a model from a log-log graph

The graph shows \(\log_{10}y\) against \(\log_{10}x\) for a model \(y=a\,x^{k}\). The line passes through (0, 0.8) and (2, 1.8). (a) Find a and k. (b) Find y when \(x=250\).

0.511.522.530.511.522.53log10 xlog10 y(0, 0.8)(2, 1.8)
Working
  1. Gradient

    \(k=\dfrac{1.8-0.8}{2-0}=0.5\)

    The gradient of a log-log line is the exponent k.

  2. Intercept

    \(\log a=0.8\), so \(a=10^{0.8}=6.31\)

    The line meets the vertical axis where log x = 0, that is x = 1.

  3. The model

    \(y=6.31\,x^{0.5}\)

  4. Predict

    \(y=6.31\times250^{0.5}=99.8\)

    Put x = 250 into the model, not into the log graph.

Video walkthrough coming