Zoho Sheet ERFC Function: Complementary Error Function and Tails
Use ERFC with symmetry controls, documented negative inputs and correctly scaled standard-normal upper-tail examples.
Verification status: See the evidence and test scope described in this guide. How formula examples are checked
ERFC returns the complementary Gauss error function, the normalized integral from its input to infinity. =ERFC(0) gives 1. For positive x it decreases toward zero; negative inputs can give values above one, as Zoho’s documented =ERFC(-1) result illustrates.
Availability and verification
Zoho’s official ERFC entry was retrieved on October 15, 2026. It documents the current rolling web service, not an original introduction release or last functioning version; both historical endpoints remain unknown. Mobile/offline parity is unverified. The additional calculations below are independent mathematical expectations, not native Zoho Sheet output.
Syntax and mathematical meaning
ERFC(number)
The one required numeric argument is the lower integration limit. No upper-limit or optional cumulative flag is documented. Mathematically, erfc(x) = (2/√π) times the integral of exp(−t²) from x to infinity. Its complement is the standard mathematical error function erf(x), so erfc(x)+erf(x)=1.
That identity is mathematical notation, not an instruction to depend on a particular product’s optional-argument ERF syntax. ERFC itself has a single unambiguous argument. Its raw output is not automatically a standard-normal tail probability; a scale change and a factor of one half are needed for that application below.
Recreate the vendor’s three controls
On ErfcControls, put the numeric inputs 0.43,−1,0 in A2:A4 and fill B2 =ERFC(A2) through B4. Expected values are approximately 0.543113305,1.842700793,1, matching the official examples.
The negative-input result is valid under the documented example; do not reject it merely because it exceeds one. The normalized erfc range for finite real x is mathematically between zero and two, unlike a probability constrained between zero and one.
A complete symmetry table
Create ErfcValues with x, Direct, Reflected and Total in A1:D1. Enter the x values below in A2:A9 and fill the following formulas through row 9:
B2: =ERFC(A2)
C2: =ERFC(-A2)
D2: =B2+C2
| Row | A: x | B: expected ERFC |
|---|---|---|
| 2 | -2 | 1.995322265 |
| 3 | -1 | 1.842700793 |
| 4 | 0 | 1.000000000 |
| 5 | 0.43 | 0.543113305 |
| 6 | 0.5 | 0.479500122 |
| 7 | 1 | 0.157299207 |
| 8 | 2 | 0.004677735 |
| 9 | 3 | 0.000022090 |
D is mathematically 2 in each row because erfc(−x)=2−erfc(x). The outputs are not odd or even; the reflected values add to two. At zero, B and C are both one.
In a copy change A6 from 0.5 to −0.5. B6 changes from approximately 0.479500122 to 1.520499878; C6 changes in the opposite direction and D6 remains two. A small tolerance is more suitable than exact floating-point equality for this identity.
Advanced: a standard-normal upper tail
For a specified standard-normal model, the probability above z is erfc(z/√2)/2. Create NormalTails with z in A1 and Upper_tail in B1. Enter z=0,1,2,3 in A2:A5 and fill B2 =ERFC(A2/SQRT(2))/2 through B5.
| Row | A: z | B: expected upper-tail probability |
|---|---|---|
| 2 | 0 | 0.500000000 |
| 3 | 1 | 0.158655254 |
| 4 | 2 | 0.022750132 |
| 5 | 3 | 0.001349898 |
The factor one half makes the z=0 tail 0.5. Without it, ERFC(0) would give one, the wrong normal-tail value. For z=−1, the upper tail is approximately 0.841344746; the direction is still “above z,” not an absolute-value tail.
Advanced: convert an original-unit threshold
Suppose a hypothetical normal model has mean 100 and standard deviation 15. For threshold 130, enter the mean in E2, standard deviation in E3 and threshold in E4. Use =ERFC((E4-E2)/(E3*SQRT(2)))/2. The standardized threshold is 2, and the expected upper-tail probability is approximately 0.022750132.
This calculation assumes a normal model and positive standard deviation; ERFC does not determine whether actual observations follow it. A probability for a continuous interval depends on both boundaries, and a density value is a different quantity again.
Small tails and numerical caution
Computing a complement by subtracting a value extremely close to one can lose significant digits. A dedicated complementary function expresses the tail directly. This is a numerical design reason to consider ERFC, not a measured guarantee that Zoho uses a particular algorithm or achieves a specified extreme-tail accuracy.
For orientation, the independent standard-normal upper tail at z=8 is about 6.220960574×10^-16. A worksheet number format with only nine decimal places would display zero even though the mathematical value is positive. Inspect scientific notation and preserve the distinction between rounding, underflow and a truly impossible event. No native extreme-tail check was performed.
Errors and troubleshooting
Zoho lists the shared invalid-name, invalid-type and invalid-reference errors, but the checked page does not define coercion for blanks, numeric strings or arrays, nor exact overflow/underflow boundaries. Use a simple documented literal to isolate a formula-entry problem before inspecting cell types.
If a supposed probability exceeds one, check whether the ERFC value was left unscaled. If a tail goes in the wrong direction, check the sign of the standardized threshold. Similar function names in other products do not prove matching precision or error handling; test transferred workbooks in their actual application.