BESSELK in Excel: Compare Modified Bessel Decay and Higher Orders

• Introduced: the first introduction version is not established by the reviewed official history.
• Availability: official help lists Excel for Microsoft 365, Excel for Microsoft 365 for Mac, Excel 2024, Excel 2024 for Mac, Excel 2021, Excel 2021 for Mac, Excel 2019, Excel 2016.
• Checked: October 14, 2026; official BESSELK reference. Documentation and independent models; no native Excel execution or wider platform certification.

=BESSELK(x,n)

Both arguments are required. X is the evaluation point, while n selects the order. Microsoft says noninteger n is truncated.

The reference’s equation image expresses K through a combination of J and Y at an imaginary mathematical argument. This is a mathematical relationship, not evidence that the worksheet J or Y functions accept complex-text inputs.

All worked x values here are strictly positive. The dedicated worksheet remarks do not provide a complete x-domain error table, so native behavior at zero or negative x is not filled in from another spreadsheet product.

Enter 0.5, 1, 2 and 3 in A2:A5. Put orders zero, one and two in B1:D1. In B2 enter =BESSELK($A2,B$1) and fill across and down.

These are independent mathematical expectations rounded to six decimals. They are not native Excel result captures.

Across each fixed-order column, the sampled values decrease as positive x increases. Within a row, higher order produces larger values in this fixture. Changing the order and changing the evaluation point are different edits with different effects.

At x=2, orders 0.9, 1.9 and 2.9 select integer orders zero, one and two under Microsoft’s truncation rule. Their independent targets are approximately 0.113894, 0.139866 and 0.253760.

If a worksheet policy requires a whole nonnegative order, validate that condition separately for numeric data rather than relying on the function to reject fractions.

The x argument is not subject to this order rule. For example, x=1.5 with order one asks for a different point than x=1 or x=2.

For positive x, an independent recurrence gives:

=BESSELK(A2,0)+2/A2*BESSELK(A2,1)

The mathematical targets match the order-two column. At x=2, the reconstruction is approximately 0.253760.

The plus sign distinguishes this recurrence from the modified I recurrence. Copying a formula from another Bessel family without checking its sign would calculate the wrong quantity.

The formula divides by x, so this audit deliberately excludes zero. A recurrence is a useful consistency check for moderate values, not a guarantee of stability for every direction of order propagation.

For order one, divide every calculated value by the value at x=1. The row at x=1 becomes one by construction.

For x=2, use:

=BESSELK(2,1)/BESSELK(1,1)

The independent ratio is approximately 0.232371. This describes the value relative to the chosen reference, not a subtraction or a percentage-point difference.

Keep the reference argument and order visible in labelled cells when extending the calculation. A denominator from a different order would create a different normalization.

For positive x, a separate mathematical identity relates modified I and K:

=BESSELI(A2,0)*BESSELK(A2,1)+BESSELI(A2,1)*BESSELK(A2,0)

The independent target is 1/A2. For the four primary x values, those targets are 2, 1, 0.5 and approximately 0.333333333.

BESSELI supplies the modified I values. This identity uses products from two distinct families; it does not identify them as aliases.

Compare unrounded calculated values using an explicitly declared tolerance appropriate to the fixture. Do not reconstruct the identity from six-decimal display values and then demand exact equality.

Microsoft assigns #VALUE! to nonnumeric x or n and #NUM! to negative n. Its dedicated worksheet page does not explicitly enumerate zero/negative-x errors, blank or logical conversion, or extreme-order behavior.

The published BESSELK(1.5,1) result is 0.277387804. Two independent mathematical implementations give approximately 0.277387800457. Both round to 0.277388 at six decimals, but the later digits differ by about 3.54×10^-9.

This difference is preserved as a precision caution. It is not labeled a proven native defect, and no unexecuted Excel accuracy guarantee is made. New examples remain independent mathematical models with positive arguments, and source equations are reviewed separately from any claim about native complex-input support.

• Microsoft BESSELK reference.
• Editorial testing standards.

Expected outputs are independently calculated examples or documented rules, not captured native Excel results. Formulas use English function names and comma argument separators. Locale, rendering and boundary restrictions are stated where relevant.