How big is South Africa, really?
Most of us learned the shape of the world from a Mercator map, and most web maps still use a version of it. Mercator was designed for navigation, not for comparing countries, so it enlarges areas increasingly towards the poles. Because much of Africa lies close to the equator, it receives far less enlargement than Europe or North America. The result is a misleading impression of their relative sizes.
On a Mercator map, the same South African outline appears almost twice as large when moved from its home latitude towards Germany. Nothing about the country has grown. The projection treats its new latitude differently.
For Day 22 I wanted to show that distortion rather than just state it. The report, The True Size of South Africa, rotates South Africa's outline across a spherical Earth towards Europe, the United Kingdom or the United States. Two synchronised maps show each step: Mercator on the left and a globe on the right. The underlying geographic area is preserved by the rotation, even though the outline changes its apparent size on Mercator.
Explore the interactive report
The page has two maps side by side, both driven by the same controls. Start at frame 0, then advance a few steps at a time and compare the active outline with its original position.
- Choose a destination: Europe (centred over Germany), the United Kingdom or the contiguous United States.
- Journey step: drag the slider from 0 (home) to 20 (arrived). South Africa moves along a great-circle path in 20 equal steps.
- Read the outlines: orange is the current position, grey shows earlier positions every four steps, and the starting position gets a white border once the outline has moved. The destination countries are blue-grey.
- Compare the maps: the Mercator view makes the effect of latitude easier to spot. The globe shows the same rotated geometry on the sphere, though its screen size also depends on the viewing angle.
- Check the numbers: a card shows the calculated area of the outline at the current step, and a table compares South Africa with the countries for the chosen destination.
Interactive report: Day 22, Projections
Neither map is an equal-area flat projection. A globe view avoids Mercator's latitude-dependent stretching but still has foreshortening towards the edge of the visible hemisphere. The KPI reports a calculated geographic area in km², not the area occupied by the shape on your screen.
Starting with Natural Earth's country outlines
All the outlines come from Natural Earth's Admin 0 Countries at 1:50m, a public-domain dataset that is detailed enough to recognise a coastline but light enough to animate. The script downloads the GeoJSON version from Natural Earth's GitHub repository: one 3.1 MB file covering every country.
I selected the main landmass of each country. For South Africa, that excludes the Prince Edward Islands, about 1,800 km to the south-east, so they do not affect the calculation of the outline's starting centre. Lesotho remains a hole in the polygon, which also helps reveal how the outline rotates. The comparisons use metropolitan France (excluding French Guiana), Germany, Italy, the United Kingdom and the contiguous United States (excluding Alaska and Hawaii). These are comparisons of the selected outlines, not necessarily every offshore territory.
From a fixed outline to a journey across the globe
Rotating, not shifting
A tempting shortcut is to add a fixed number of degrees to each vertex's latitude and longitude. But degrees are not uniform distances: a degree of longitude spans less ground as latitude increases. Such a shift would change the physical geometry instead of merely repositioning it.
Instead, the Python script converts each boundary vertex to a three-dimensional point on a sphere and applies Rodrigues' rotation formula about an axis through the Earth's centre. That moves the outline's chosen centre along a great-circle route. A rigid spherical rotation preserves the underlying distances and area; any small variation in the reported area comes from the method used to calculate it afterwards.
Each journey is split into 20 equal steps:
- Europe (over Germany): about 9,000 km, roughly 450 km per step.
- United Kingdom: about 9,600 km, roughly 480 km per step.
- United States: about 14,800 km, roughly 740 km per step.
The outline turns as it travels
One thing surprised me. On the journey to the USA, South Africa arrives visibly rotated relative to north. That isn't a bug. A great-circle path changes compass bearing as it goes, and the outline turns with it. The Europe journey runs nearly north-south, so the effect there is small. The USA journey crosses the Atlantic diagonally, so it's obvious.
Calculating the area, and the small drift
The script estimates each frame's geographic area using a longitude–latitude ring-area summation on a sphere with mean radius 6,371.0088 km. It subtracts interior holes, including Lesotho, from the outer boundary. At frame 0, the selected South African outline gives 1,221,410 km², close to the commonly quoted figure of about 1.22 million km². This is a modelled area from the Natural Earth geometry, not a cadastral or survey measurement.
The resulting figures drift slightly over the 21 frames: a range of 100 km² for Europe, 112 km² for the UK and 74 km² for the USA, all under 0.01% of the starting area. This is not evidence that the country has changed size. The longitude–latitude edge summation is an approximation whose result is not perfectly invariant under rotation; rounding contributes far less than the treatment of those edges. The KPI deliberately shows this small calculation drift rather than replacing it with a fixed reference number.
The geometry generation uses a Python script and only the standard library. It produces 63 journey rows (21 for each of three destinations) and a separate file with five comparison-country outlines. The report's prepared map table has 168 rows, as the comparison outlines are repeated for each journey frame. Keeping the shapes in WKT means Icon Map Pro can render them from Power BI data.
Two projections from one Icon Map Pro setup
Same data, different projection
Both maps are Icon Map Pro visuals bound to the same map table and measures. One uses Mercator and the other a globe view. They respond to the same destination and frame controls, displaying the same rotated WKT geometry. They are not meant to produce identical-looking shapes in screen pixels: the projection, zoom and globe viewpoint each influence the rendering.
One measure decides which outlines to draw
Drawing all 21 frames at once made the map a blur, so a DAX measure decides which outlines to show. It finds the latest step selected on the slider and returns the WKT for:
- the current step;
- earlier steps divisible by four (0, 4, 8, 12, 16), as a sparse trail;
- the comparison countries, always.
Everything else returns blank, and Icon Map Pro does not draw it. Separate DAX measures control fill and stroke colours. Frame 0 is orange while it is current; once the journey advances, its white border marks the starting location, with intervening sampled positions shown in grey.
A slider that keeps the history
To make the trail work, the step slicer needs to select every frame from 0 up to the chosen step, not just one. Setting it to Less than or equal to gives a single handle that does exactly that. The measure then uses the highest selected step as the current position.
The comparison-country shapes are present at each frame, so the step slicer does not make the destination reference disappear.
A slicer that wasn't filtering
During development, the comparison table initially continued to show Europe even when another destination was selected. The destination slicer's interaction with that visual had been set to None, so the measure did not receive the intended filter context. Changing the interaction to Filter resolved it. In Power BI, checking Edit interactions is as important as checking the DAX when a measure appears to ignore a slicer.
Designing for a dark, quiet basemap
Both maps use Icon Map's DataViz Black basemap: dark, low-contrast, with thin dashed borders. It lets the outlines carry the story.
- Current position: orange fill (#F28C28) with an amber outline, the strongest colour on the page.
- Earlier positions: grey (#8A8F98), present but quiet.
- Starting position: a white border, so you can always see where the journey began.
- Destination countries: blue-grey (#536578), clearly different from the moving shape.
The polygon fills use high transparency so overlapping outlines remain readable. A compact text legend above each map explains the four roles without taking space away from the maps.
What the map reveals
1. South Africa appears almost twice as large on Mercator when moved towards Germany. The local area-enlargement factor in a spherical Mercator projection is approximately sec²(latitude). At South Africa's starting centre (about 29°S), it is around 1.3; at the destination over Germany (51°N), around 2.5. That suggests an apparent-area increase of roughly 1.9 times for the same outline. The corresponding centre-latitude estimates are about 2.2 times over the UK and 1.3 times over the central USA. These are approximate projection-scale comparisons, not measured changes in the number of pixels covered by the complete polygon; the scale varies from one part of the outline to another.
2. France, Germany and Italy together are almost the same area as South Africa. In the selected Natural Earth outlines, their combined area is 1,202,179 km², or 98.4% of South Africa's 1,221,410 km². When viewed in their original locations on Mercator, the European countries benefit from much greater latitude-related enlargement. Moving South Africa north towards Germany puts the outlines at more comparable latitudes, making a visual comparison more meaningful. The shapes do not fit neatly inside one another, and the two map views may look more similar than different at the destination. The numerical areas provide the firmer comparison.
3. South Africa is approximately five times the size of the United Kingdom. The selected UK outline covers 239,611 km², making South Africa about 5.1 times larger by area. Moving South Africa over the UK puts the outlines in a similar latitude band, so Mercator no longer gives one country such a large latitude-related advantage over the other. This makes the visual comparison easier to judge, although the measured figures remain the best guide.
4. The contiguous United States is around 6.5 times the size of South Africa. Its selected mainland outline covers 7,933,175 km², about 6.5 times South Africa's area. The USA journey is also a useful demonstration of orientation: the South African outline turns noticeably as it follows the great-circle route across the Atlantic. It is the same rotated geometry, not a new shape or a simple shift in longitude.
The central lesson: The most striking contrast is often South Africa's starting versus destination position on Mercator, rather than an obvious difference between the Mercator and globe views at the same step. The globe is a useful check on the spherical movement, but it introduces its own perspective and foreshortening. For decisions that depend on area, compare the numeric measurements or use an appropriate equal-area projection.
What I learned from Day 22
- Move shapes on the sphere, not the grid. Adding degrees to coordinates distorts the shape; rotating about the Earth's centre keeps it true.
- Expect the outline to turn. Great-circle journeys change bearing, so the shape rotates relative to north. It's correct, and worth explaining.
- Separate calculation-method drift from real change. The area KPI varies by less than 0.01% because of the chosen geographic area approximation, not because a rigid rotation changes the country.
- Show less, so the current position stands out. A sparse trail every four steps tells the story better than 21 overlapping outlines.
- Check slicer interactions when a measure reads an unrelated table. The DAX can be right, and the visual still never sees the filter.
Final thoughts
Business intelligence reports put countries, regions and sales territories on maps all the time, and the reader's eye compares their sizes whether we intend it or not. Consider a Power BI report comparing equally sized sales territories in northern Europe and southern Africa. The northern territory can appear substantially larger on a Mercator map, even when the underlying business measures tell a different story. The map has not changed the data, but it can influence the reader's interpretation.
On a Mercator-style web map, apparent geographic areas are affected by latitude. The practical response is to display measured areas beside the map, or choose an equal-area projection when area is central to the question. For this challenge, moving one familiar outline across the Earth makes that lesson visible.
South Africa didn't grow as it travelled north. The map made it look that way.
Sources and attribution
- Country outlines: Natural Earth, Admin 0 Countries, 1:50m, public domain. Made with Natural Earth (https://www.naturalearthdata.com/downloads/50m-cultural-vectors/50m-admin-0-countries-2/). GeoJSON from the natural-earth-vector repository (https://github.com/nvkelso/natural-earth-vector), retrieved 19 November 2026.
- Basemap: Icon Map DataViz Black style, © Tekantis, Protomaps and OpenStreetMap contributors.
- Mapping: Icon Map Pro by Tekantis, in Microsoft Power BI.
- Projection background: US Geological Survey, Map Projections (https://pubs.usgs.gov/gip/70047422/report.pdf), for the properties and limitations of Mercator.
- Area-method context: D3, Spherical Math (https://d3js.org/d3-geo/math), describes a different spherical-area method; the figures in this article come from the project's own Python approximation, not D3.
Methodology note: The selected mainland outline of South Africa (Natural Earth 1:50m; Prince Edward Islands excluded and Lesotho retained as an interior hole) is rotated as a set of three-dimensional unit vectors using Rodrigues' rotation formula. The starting centre is estimated from the source outline in geographic coordinates and moved along a great circle to 51.0°N, 10.4°E (Europe), 54.0°N, 2.5°W (UK) or 39.5°N, 98.5°W (contiguous USA), in 20 equal angular steps. The script estimates area from the longitude–latitude rings using a spherical edge summation and an Earth radius of 6,371.0088 km; it does not implement an exact spherical-excess algorithm. The observed drift of up to 112 km² is a property of that approximation and the rounded output, not a physical area change. Comparison countries are subsets from the same Natural Earth dataset and are processed with the same area method. Mercator area-enlargement factors use sec²(latitude) at the selected outline centre; they are indicative local factors, not whole-polygon pixel-area measurements. The globe view is a perspective rendering, not an equal-area projection.