Navigate
Roadmap Tutorials Exercises Tools Get certified

Fix 'could not find function' in R

The error could not find function "X" means R searched your workspace and every loaded package and found no function with that name. The usual cause is that the package owning the function was never loaded with library(), followed by plain typos.

By Selva Prabhakaran  ·  Published July 22, 2026  ·  Last updated July 22, 2026
⚡ Quick Answer
library(dplyr)                  # the usual fix: load the owning package
install.packages("dplyr")       # one-time install if loading fails
exists("select")                # FALSE means R cannot see the name
find("mean")                    # which package owns a visible function
dplyr::select(mtcars, mpg)      # call one function without attaching
sessionInfo()                   # list the packages currently loaded

Need explanation? Read on for examples and pitfalls.

What this error means

R resolves every function name at the moment you call it. When you type select(mtcars, mpg), R walks its search path: first your workspace, then each attached package in order, then base R. If no environment on that path defines select, the call stops with this error. The quoted name in the message is exactly what R looked for, spelling and capitalization included.

The shortest way to trigger it is to call a package function in a fresh session:

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.
RReproduce the error
select(mtcars, mpg) #> Error in select(mtcars, mpg) : could not find function "select"

  

  





select() lives in the dplyr package. A fresh R session attaches only the base packages, so the name does not exist yet, no matter how many times you installed dplyr in the past.



Key Insight
Installed and loaded are different states. install.packages() copies files to disk once per machine. library() puts those functions on the search path, once per session. This error is almost always about loading, not installation.



Diagnose it in 30 seconds



Two base functions locate the problem. exists() says whether R can see the name at all, and find() reports which attached package owns it:




  

    
RCheck what R can see

    
exists("select")   # can R see the name right now?
#> [1] FALSE

exists("mean")     # a base function, for comparison
#> [1] TRUE

find("mean")       # which package owns it
#> [1] "package:base"

    

      
      
    

    

  

  





Match what you observe against this table:







Symptom
Likely cause
Fix






exists("fn") is FALSE and you never ran library()
Package not loaded this session
library(pkg) at the top of the script




library(pkg) itself fails with "there is no package"
Package not installed
install.packages("pkg") once




The quoted name looks almost right
Typo or wrong capitalization
Match the documented spelling exactly




Works in your console, fails on knit or Rscript
The fresh session never loads the package
Put library() calls inside the document




The message names %>%
Pipe package not attached
library(magrittr) or switch to the native pipe







Tip
When you do not know the owning package, ask R. ??select searches the documentation of every installed package, and getAnywhere("select") inspects loaded namespaces even when they are not attached.



The four causes and their fixes



1. The package is not loaded



This is the cause in most real sessions. The function exists on disk, but the current session never attached its package. Load it, and the same call works:




  

    
RLoad the package then call

    
library(dplyr)

head(select(mtcars, mpg), 3)
#>                mpg
#> Mazda RX4     21.0
#> Mazda RX4 Wag 21.0
#> Datsun 710    22.8

    

      
      
    

    

  

  





Every script should load its packages at the top, one library() call per line. The packages chapter of the official An Introduction to R manual explains the attach mechanism in depth.



2. The package is not installed



Loading fails when there is nothing to load. If library(dplyr) answers with "there is no package called 'dplyr'", the package never made it onto this machine. Install it once, then load it every session:




  

    
RInstall once then load

    
# install.packages("dplyr")   # run once per machine
library(dplyr)                # run in every session

    

      
      
    

    

  

  





This split trips people moving to a new computer or a fresh R upgrade: scripts full of library() calls fail until the packages are reinstalled.



3. A typo or wrong capitalization



R is case-sensitive, so Mean and mean are different names. The error message quotes the exact string R searched for, which makes typos easy to spot once you know to look:




  

    
RFix the capitalization

    
Mean(c(2, 4, 9))
#> Error in Mean(c(2, 4, 9)) : could not find function "Mean"

mean(c(2, 4, 9))
#> [1] 5

    

      
      
    

    

  

  





Read the quoted name character by character. lenght(), summarise() versus summarize(), and stray autocomplete artifacts account for a surprising share of these errors.



4. The function is newer than your package version



A current tutorial plus an old installation produces a missing function. Packages gain functions over time; dplyr's reframe(), for example, only exists from version 1.1.0 onward. Check what you have:




  

    
RCheck the installed version

    
packageVersion("dplyr")
#> [1] '1.2.1'

    

      
      
    

    

  

  





If your version predates the function you need, update.packages() or a fresh install.packages("dplyr") brings it in. The same applies to base R itself: code using the native pipe fails on R older than 4.1.0.



Warning
Knitting runs in a fresh session. R Markdown and Quarto render documents in a new R process, so packages loaded in your console do not carry over. Code that works interactively still fails at render time unless the document itself calls library() in an early chunk.



The %>% pipe case



The pipe is a function too, and it lives in a package. %>% comes from magrittr and is re-exported by dplyr and the rest of the tidyverse. Pasting piped code into a session with no packages loaded fails immediately:




  

    
RThe pipe needs its package

    
# In a fresh session, mtcars %>% nrow() stops with:
# Error in mtcars %>% nrow() : could not find function "%>%"

library(magrittr)   # or library(dplyr), which re-exports the pipe
mtcars %>% nrow()
#> [1] 32

    

      
      
    

    

  

  





Since R 4.1.0 there is also the native pipe |>, which is part of base R and needs no package at all. If you keep hitting this error in short scripts, the native pipe removes the dependency.



How to avoid it next time



A few habits make this error rare.



    


  • Start every script with its library() calls, grouped in one block, so a fresh session can run it top to bottom.
    • 


  • Use pkg::fun() for one-off calls, such as dplyr::select(). The :: form works without attaching and documents where the function comes from.
    • 


  • Restart R regularly and rerun from the top. A long-lived session hides missing library() lines until the worst moment.
    • 


  • After following a tutorial, check versions with packageVersion() before assuming your code is wrong.
    • 





Note
Coming from Python? This is R's equivalent of NameError: name 'x' is not defined after a missing import. The fix is the same idea: load the module, or qualify the call with its package name.



Try it yourself





Try it: The broken line below throws the missing-function error because of capitalization. Write the working call so ex_total holds the sum of 12, 6, and 9.



  

    
RYour turn: fix the missing function

    
# Sum(c(12, 6, 9)) fails: could not find function "Sum"
ex_total <- # your code here

ex_total
#> Expected: [1] 27

    

      
      
    

    

  

  





Click to reveal solution



  

    
RSolution

    
ex_total <- sum(c(12, 6, 9))
ex_total
#> [1] 27

    

      
      
    

    

  

  





Explanation: R is case-sensitive, so Sum does not match the base function sum. Lowercasing the name lets R find it on the search path.








FAQ



Why is R saying "could not find function %>%"?



Because the magrittr pipe was never attached in the current session. %>% is an operator function from the magrittr package, re-exported by dplyr and tidyverse. Run library(dplyr) or library(magrittr) before the piped code. On R 4.1.0 or newer you can also replace %>% with the native pipe |>, which belongs to base R and never triggers this error.



Why is my object not being found in R?



That is the sibling message, object 'x' not found, and it fires when a name is used as a value rather than called as a function. The line that creates the object never ran, ran with an error, or spelled the name differently. Rerun the creating line, then check with exists("x"). The object not found error guide walks through each case.



Is there a find function in R?



Yes. find("mean") returns which attached package defines a name, and exists("mean") answers whether the name is visible at all. For functions in installed but unattached packages, getAnywhere("fn") searches loaded namespaces, and ??fn searches the help index of everything installed. Together they answer both halves of this error: does the function exist, and where does it live.



What is the missing function in R?



In this error, the missing function is whatever name appears in quotes: R searched the workspace and all attached packages and found no function with that name. Separately, base R has a function called missing(), which tests inside a function whether a caller supplied an argument. The two are unrelated; missing() is about arguments, while this error is about name lookup.





Start with the parent guide to common R errors and how to read them. This error has a mirror: parentheses on a data object report a missing function, while brackets on a function report object of type closure is not subsettable. For the mechanics behind the fixes, see installing and loading packages, the pipe operator in R, and R environments for how the search path resolves names.