Navigate
Home Compendium
Exercises
Tidyverse packages
dplyr ggplot2 tidyr stringr lubridate purrr forcats readr broom Tidyverse R for Data Science
Deep dives
dplyr Joins dplyr Window Functions dplyr group_by ggplot2 Themes ggplot2 Facets ggplot2 Color Scales ggplot2 Bar Charts ggplot2 Heatmaps tidyr Pivot tidyr Nest/Unnest Regex
Wrangling & EDA
Data Wrangling Data Cleaning EDA Data Visualization data.table dbplyr / SQL Web Scraping API Calls
Statistics
Hypothesis Testing Linear Regression Logistic Regression Correlation Probability Distributions
Machine Learning
Machine Learning tidymodels Random Forest XGBoost Clustering PCA Cross-Validation
Time Series
Time Series ARIMA
By Industry
Finance Marketing Analytics Healthcare Biostatistics Genomics Sports Analytics A/B Testing
Reporting & Apps
Shiny R Markdown
Levels
R for Beginners R Interview Questions
R r‑statistics.co

ggplot2 coord_fixed() in R: Lock Plot Aspect Ratio

ggplot2 coord_fixed() in R locks the aspect ratio between x and y, so one data unit on x always takes the same screen space as one data unit on y. Use it whenever the angle of a slope, a distance, or a shape carries meaning, including maps, predicted-vs-actual diagnostics, and y = x comparisons.

By Selva Prabhakaran  ·  Published May 15, 2026  ·  Last updated May 15, 2026
⚡ Quick Answer
coord_fixed()                                          # default 1:1 ratio
coord_fixed(ratio = 1)                                 # explicit 1:1
coord_fixed(ratio = 2)                                 # y unit twice the x unit
coord_fixed(ratio = 0.5)                               # x unit twice the y unit
coord_fixed(xlim = c(0, 10), ylim = c(0, 10))          # fixed ratio plus zoom
coord_fixed(expand = FALSE)                            # remove default padding
ggplot(df, aes(x, y)) + geom_point() + coord_fixed()
ggplot(df, aes(long, lat)) + geom_polygon() + coord_fixed(1.3)

Need explanation? Read on for examples and pitfalls.

📊 Is coord_fixed() the right tool?
STARTenforce 11 visual scale on x and y: coord_fixed()zoom without dropping datacoord_cartesian(xlim = c(a, b))project longitude and latitude properlycoord_sf() or coord_map()rotate the plot 90 degreescoord_flip()bend an axis into a circlecoord_polar()control output figure size, not data ratioggsave(width = w, height = h)log or sqrt transform an axisscale_x_log10(), scale_x_sqrt()

What coord_fixed() does in one sentence

coord_fixed() pins the visual y-to-x ratio so equal data distances render as equal screen distances. Without it, ggplot2 stretches the panel to fill the available space, which means a 45 degree line on the data scale almost never looks 45 degrees on screen. With coord_fixed(ratio = 1), one y unit renders the same pixels as one x unit, regardless of panel size.

The ratio argument is y / x. A value of 2 makes a y unit twice as tall as an x unit; 0.5 makes it half as tall.

Syntax

coord_fixed() takes five arguments and zero are required.

Run live
Run live, no install needed. Every R block on this page runs in your browser. Click Run, edit the code, re-run instantly. No setup.
Rcoord_fixed function signature
coord_fixed( ratio = 1, # y / x screen ratio; 1 means equal xlim = NULL, # optional zoom window on the x axis ylim = NULL, # optional zoom window on the y axis expand = TRUE, # add 5 percent padding around the limits clip = "on" # "on" trims geoms at the panel; "off" lets them spill )

  

  





    


  • ratio: a positive number. Defaults to 1, the most common case. Pass ratio = 1.3 for unprojected lat-long data at mid latitudes.
    • 


  • xlim and ylim: behave the same as in coord_cartesian(); applied after stats run, so smoothers see all data.
    • 


  • expand: set FALSE to anchor a bar baseline or crop a map to its exact extent.
    • 


  • clip: switch to "off" for geom_text() labels that should render outside the panel.
    • 





Note
coord_equal() is the same function with a different name. coord_equal(ratio = 1) and coord_fixed(ratio = 1) are identical; coord_equal() is a thin wrapper kept for historical reasons. Pick one and stay consistent across your codebase.



Five common patterns



Pattern 1: square scatter for predicted vs actual. Diagnostic plots compare predicted to actual along the y = x line. A wide panel tilts the line and makes overestimation look like underestimation. Lock the ratio so the reference line is a true 45 degrees.




  

    
RPredicted vs actual with a true 45 degree line

    
library(ggplot2)

fit <- lm(mpg ~ wt + hp, data = mtcars)
diag <- data.frame(actual = mtcars$mpg, predicted = fitted(fit))

ggplot(diag, aes(predicted, actual)) +
  geom_point() +
  geom_abline(slope = 1, intercept = 0, color = "red") +
  coord_fixed() +
  labs(x = "Predicted mpg", y = "Actual mpg")
#> Square panel; the red line bisects the panel at a true 45 degrees.

    

      
      
    

    

  

  





Pattern 2: map with one degree of latitude equal to one degree of longitude. For an unprojected map at the equator, set ratio = 1. At mid latitudes, set ratio = 1 / cos(lat) to approximate the Mercator squeeze; at 40 degrees that is roughly 1.3.




  

    
RState map with equator-style 1:1 ratio

    
states <- map_data("state")

ggplot(states, aes(long, lat, group = group)) +
  geom_polygon(fill = "lightgray", color = "white") +
  coord_fixed(ratio = 1.3) +
  labs(x = NULL, y = NULL)
#> Map of the lower 48; states no longer squeezed horizontally.

    

      
      
    

    

  

  





Pattern 3: shape preservation for ellipses and circles. When you draw a circle or ellipse with geom_path(), the panel must use equal scaling or the circle becomes an oval. coord_fixed() guarantees that a circle of radius r looks like a circle.




  

    
RDraw a true circle

    
theta <- seq(0, 2 * pi, length.out = 200)
circle <- data.frame(x = cos(theta), y = sin(theta))

ggplot(circle, aes(x, y)) +
  geom_path() +
  coord_fixed() +
  labs(title = "Unit circle, drawn as a circle")
#> The path closes into a true circle, not an ellipse.

    

      
      
    

    

  

  





Pattern 4: residual plots where slope angle matters. A residuals-vs-fitted plot reads better when residuals and fitted values share the same visual scale; trends appear at their honest steepness instead of being exaggerated by the panel.




  

    
RResiduals with truthful slopes

    
fit <- lm(mpg ~ wt, data = mtcars)
res <- data.frame(fitted = fitted(fit), resid = resid(fit))

ggplot(res, aes(fitted, resid)) +
  geom_point() +
  geom_hline(yintercept = 0, color = "blue") +
  coord_fixed() +
  labs(x = "Fitted mpg", y = "Residual")
#> Tall, narrow panel; residual scatter looks faithful to its true magnitude.

    

      
      
    

    

  

  





Pattern 5: custom ratio to emphasize a small range. Setting ratio = 5 makes a y unit five times taller than an x unit, useful when the meaningful variation in y is small relative to the x range and you want to amplify it without rescaling the data.




  

    
RAmplify a narrow y range

    
df <- data.frame(x = 1:50, y = 0.5 + sin(1:50 / 5) * 0.1)

ggplot(df, aes(x, y)) +
  geom_line() +
  coord_fixed(ratio = 50) +
  labs(x = "Index", y = "Value")
#> Wavelets become clearly visible without changing the underlying numbers.

    

      
      
    

    

  

  





Tip
Pick coord_fixed() any time slope, distance, or shape carries meaning. Maps, residuals, predicted vs actual, geometry plots, force diagrams, and dose-response curves all need it. Skip it for time series, bar charts, and most categorical comparisons; the panel-stretching default is fine when only the bar heights or trend direction matter.



coord_fixed() vs coord_cartesian() vs theme(aspect.ratio)



Three ways to control plot shape, but they answer different questions.







Feature
coord_fixed()
coord_cartesian()
theme(aspect.ratio = 1)






Locks data unit ratio (1 unit y = 1 unit x)
yes
no
no




Locks panel pixel ratio (height / width)
depends on data range
no
yes




Affects map and circle correctness
yes
no
no




Honors limits without dropping data
yes
yes
no, theme is layout only




Use case
maps, residuals, y=x
zoom panel only
uniform panel shape across facets







Key Insight
coord_fixed() is about data units; theme(aspect.ratio) is about panel pixels. A coord_fixed() plot of x in [0, 10] and y in [0, 100] produces a panel ten times taller than wide, because a y unit equals an x unit and there are ten times more y units. A theme(aspect.ratio = 1) plot always renders a square panel, regardless of the data range, so a unit on x rarely matches a unit on y. Use coord_fixed() when the geometry must be honest. Use aspect.ratio when you want every facet to share a shape.



Common pitfalls



Pitfall 1: tiny panels when y range is much smaller than x range. coord_fixed() with x in [0, 1000] and y in [0, 5] produces a panel 200 times wider than tall, which often collapses to a thin strip. Either pass a custom ratio to widen the y unit, or rescale one of the variables before plotting.



Pitfall 2: applying coord_fixed() to a faceted plot and getting blank panels. Each facet inherits the same data-unit ratio, but free scales (facet_wrap(scales = "free")) can produce panels with wildly different aspect ratios. Stick to scales = "fixed" when using coord_fixed() with facets, or drop coord_fixed() and use theme(aspect.ratio = 1) for uniform panel shapes.



Pitfall 3: using coord_fixed() on a map with raw lat-long values. At latitudes far from the equator, one degree of longitude covers fewer kilometers than one degree of latitude. coord_fixed(ratio = 1) squeezes the map. Use coord_fixed(ratio = 1 / cos(mean_lat * pi / 180)), or switch to coord_sf() for any production map.



Pitfall 4: combining coord_fixed() with another coord and seeing one ignored. A ggplot2 plot has only one coord. Calling coord_polar() after coord_fixed() overwrites the fixed ratio. To zoom a fixed-ratio plot, pass xlim and ylim to coord_fixed() itself; do not chain coords.



Pitfall 5: assuming ratio = 1 is the same as theme(aspect.ratio = 1). They produce identical panels only when the x and y data ranges are equal. Otherwise, coord_fixed(1) follows the data and theme(aspect.ratio = 1) follows the layout, and the two diverge.



Try it yourself





Try it: Build a scatter plot of Sepal.Length vs Sepal.Width from the iris dataset, colored by Species, with a 1:1 aspect ratio so the visual distance between any two points reflects the true Euclidean distance. Save the plot to ex_iris.



  

    
RYour turn: equal-scale iris scatter

    
# Try it: 1:1 iris scatter
ex_iris <- # your code here

ex_iris
#> Expected: scatter colored by Species; one cm on x equals one cm on y.

    

      
      
    

    

  

  





Click to reveal solution



  

    
RSolution

    
library(ggplot2)

ex_iris <- ggplot(iris, aes(x = Sepal.Length, y = Sepal.Width, color = Species)) +
  geom_point(size = 2) +
  coord_fixed() +
  labs(x = "Sepal length (cm)", y = "Sepal width (cm)")

ex_iris
#> Three colored point clouds; panel proportions follow the data ranges.

    

      
      
    

    

  

  





Explanation: Both axes are measured in centimeters, so a 1:1 ratio is honest. Without coord_fixed(), the panel would stretch and the visual distance between two flowers would not reflect their true botanical distance.










    


  • coord_cartesian(): zoom a panel to a window without dropping data; does not lock ratio.
    • 


  • coord_equal(): identical to coord_fixed(); older alias kept for compatibility.
    • 


  • coord_sf(): proper map projections for spatial data, replacing the lat-long approximation.
    • 


  • coord_flip(): rotates the plot 90 degrees; horizontal bar charts are the typical use.
    • 


  • coord_polar(): bends an axis into a circle, used for pie and rose charts.
    • 





FAQ



What does coord_fixed() do in ggplot2?



coord_fixed() locks the aspect ratio of the plot panel so equal data distances on x and y render as equal pixel distances. The default ratio = 1 makes one unit of y the same screen size as one unit of x, which is essential for maps, predicted-vs-actual diagnostics, and any plot where slope angle or shape carries meaning. Without it, ggplot2 stretches the panel to fill available space and distorts angles.



What is the difference between coord_fixed and coord_equal?



There is no functional difference. coord_equal() is an alias for coord_fixed(); both accept identical arguments and return the same object. The ggplot2 source defines coord_equal() as a thin wrapper kept for historical naming. Pick one and stay consistent. Most modern code prefers coord_fixed() because the name documents what is being fixed (the ratio between axes).



How do I use coord_fixed for a map in ggplot2?



For a quick unprojected map of US states near mid latitudes, use coord_fixed(ratio = 1.3); the value 1.3 is roughly 1 / cos(40 degrees) and compensates for meridian convergence. For production maps or anything outside a narrow latitude band, switch to coord_sf() from the sf package for a proper projection. Avoid coord_fixed(1) on lat-long data above 30 degrees; the map will look horizontally squashed.



Can I combine coord_fixed with xlim and ylim to zoom?



Yes. coord_fixed() accepts xlim and ylim arguments that behave like in coord_cartesian(): they crop the panel without dropping data, so smoothers and densities still see every row. The aspect ratio is preserved inside the cropped window. Do not call coord_cartesian() separately to zoom; it would overwrite coord_fixed() because a plot can only have one coord.



External reference: ggplot2 coord_fixed() documentation.



          

        


        
      


      

        

        
© 2016-2026 Selva Prabhakaran.
          This work is licensed under the Creative Commons License.
        

        

          About ·
          RSS