R was started by Ross Ihaka and Robert Gentleman at the University of Auckland in New Zealand, with its first announcement usually dated to 1993. It is a free implementation of ideas from the S statistics language developed at Bell Labs. Ask why the R language was named after its creators and you get a two-part answer: R is the first letter of both Ross and Robert, and it is the letter just before S, a small joke about where the language came from.
Origins in Auckland
Ihaka and Gentleman were statistics lecturers who needed a good environment for teaching. The commercial standard was S, created by John Chambers and colleagues at Bell Labs from 1976 and sold as S-PLUS. Instead of cloning S's internals, they built R's interpreter on Scheme-like foundations: lexical scoping and closures, with S-style syntax on top. Ihaka described the early design in a 1996 paper with Gentleman in the Journal of Computational and Graphical Statistics.
Encouraged by Martin Mächler of ETH Zurich, they released the source code under the GNU General Public License in 1995. A wider group of contributors formed the R Core Team in 1997, and R 1.0.0 was released on 29 February 2000.
Why the R language was named after its creators
The name was a practical choice. Software names from S, such as S-PLUS, were trademarks, and the authors wanted something short that signalled kinship without claiming to be S. "R" did both: it pointed at Ross and Robert, and it sat alphabetically next to S. The language was compatible with much S code from early on, so the letter told statisticians exactly what to expect.
Design and key features
- Vectors everywhere: even a single number is a vector of length one, and arithmetic works element-wise without loops.
- Data frames: tables of mixed-type columns, the model later copied by pandas in Python.
- Functional core: first-class functions, closures and lazy argument evaluation (promises).
- Several object systems: S3, S4, Reference Classes, and packages such as R6.
- Graphics: publication-quality plots in base R, extended by lattice and ggplot2.
A short R example
data(mtcars)
fit <- lm(mpg ~ wt + hp, data = mtcars)
summary(fit)$r.squared
mtcars |>
subset(cyl == 4) |>
with(mean(mpg))Fitting a linear regression takes one line, and the native pipe |>, added in R 4.1, chains steps together.
CRAN, the tidyverse and RStudio
The Comprehensive R Archive Network, CRAN, started in 1997 and now hosts more than twenty thousand packages, each checked automatically across platforms. Bioconductor, launched by Gentleman in 2002, adds thousands more for genomics.
Hadley Wickham's ggplot2 (2005), based on Leland Wilkinson's The Grammar of Graphics, changed how people make charts, and his dplyr, tidyr and readr packages grew into the tidyverse. The RStudio IDE, first released in 2011 by a company founded by J. J. Allaire, made R far more approachable; that company renamed itself Posit in 2022 to reflect its support for Python as well. R Markdown and its successor Quarto made reproducible reports routine.
Versions timeline
| Year | Release | Notable change |
|---|---|---|
| 1993 | First announcement | Interpreter shown on the S-news list |
| 1995 | GPL source release | R becomes free software |
| 2000 | R 1.0.0 | First stable release |
| 2004 | R 2.0.0 | Lazy loading of data and code |
| 2013 | R 3.0.0 | Long vectors on 64-bit systems |
| 2020 | R 4.0.0 | stringsAsFactors = FALSE by default, raw strings |
Where R is used today
R is the working language of academic statistics, epidemiology, ecology, psychology and bioinformatics. Pharmaceutical companies use it in regulatory submissions, the BBC's data team has published its own R graphics package, and the Shiny framework turns analyses into interactive web apps. Python has taken much of general machine learning, but for statistical modelling and visualisation R remains the specialist's tool.
Influence and legacy
R showed that an open-source statistics environment could beat expensive commercial packages, and its data frame became the template for pandas, Spark DataFrames and Julia's DataFrames.jl. Julia's designers listed R among their influences. Knowing why the R language was named after its creators also says something about its spirit: a modest, academic project that grew into infrastructure for modern data science.
Frequently asked questions
Is R a free version of S?
R is an independent implementation that runs most S code, but it is not a copy. Its internals use Scheme-style lexical scoping, which differs from S in subtle ways. John Chambers, the creator of S, later joined the R Core Team.
Who maintains the R language today?
The R Core Team, a group of around twenty volunteers, controls the source code. The R Foundation, a non-profit founded in 2003, holds the copyright and supports the project. Posit and other companies fund many popular packages but do not own R.
Should I learn R or Python for statistics?
For classical statistics, biostatistics and visualisation, R has deeper and more specialised packages. Python is stronger for general programming and deep learning. Many analysts learn both, and tools like Quarto let you mix them in one document.
