Linked answers

Linked answers are what make a sheet behave like a small live model instead of a list of one-off calculations. Any answer can be reused elsewhere as a token that always shows the current value of its source line.

Inserting a token

Double-click any answer — or type an operator on a new line — and Numlex inserts a live token at the caret or selection. A token is a small bubble that always displays the current value of its source line: change the source and the bubble updates immediately.

What can become a token. A double-click mints a token from a number, a named value (including integer names) and a money answer. The automatic previous-answer path (typing an operator on a new line) additionally accepts an exact integer answer. Boolean, date, location and DMS answers are never tokenized. See Settings & appearance for the per-answer formatting controls that sit alongside tokens.

hotel = 1240              1,240
hotel + 10%               1,364
deposit = hotel + 10%     1,364
deposit / 4               341

Several distinct bubbles may reuse the same earlier line — each keeps its own bubble and follows the live value.

Live propagation

Because tokens are references, not copies, edits propagate:

• Change the source line and every token that uses it re-evaluates.
• If a source line stops evaluating, its tokens stay in place and show the remembered Line N label instead of a stale number.
• A line never depends on itself: inserting a token while the caret is on the source line itself puts the token on a new line just below.

Using tokens

Unitless tokens join the same engine: sqrt(<token>) and <token> ^ 2 evaluate with the token as a live argument. Function calls are scalar, so unit-bearing and money values — including tokens — do not belong inside them; use a conversion or plain arithmetic instead.

Converting a token

A token converts exactly like the value it stands for. Start a conversion line with the token and the result matches converting the source value directly:

• A token that carries a unit keeps its own unit: <token> to kg, <token> in kg.
• A unitless token names its source unit: <token> mg to kg.
• A unit-bearing token may also repeat its own unit before the keyword: <token> mg to kg. Aliases count as the same unit, so inch works for in.

4673 mg
 to kg              → 0.004673 kg
 in kg              → 0.004673 kg

4673
 mg to kg           → 0.004673 kg

in is both the inch unit and a conversion keyword, so an inch-carrying token followed by in is read as the keyword:

3 in
 in cm              → 7.62 cm
 in in              → 3 in

Errors are the same ones a typed conversion line reports. An unrecognized unit label is Unknown units; two known but different units — including another temperature scale or another currency code — are Incompatible units, and a token’s own unit is never silently replaced by the one you typed.

The conversion stays live: the value is substituted at full engine precision (max(decimalPlaces, 10)), not rounded for display, and editing the source line changes the converted answer like any other token.

Per-answer menu

Right-click (or Control-click) any answer for its native menu:

• Copy Answer — puts the exact displayed value on the clipboard.
• A 0…10 dp slider — re-rounds just that answer without touching the source line, independently of the global rounding set on the Numbers tab.
• A Number Format submenu — a display-only choice for this answer; a local display override never changes the value, and tokens that reference the line keep their own or the global format (see Settings & appearance).
• Delete Line — removes the source line.

The menu is fully native, so it follows the system appearance in Light and Dark. It exists only for answers that have one, and it contains no highlight item: line highlights live under Format → Highlight and the editor’s text context menu, and they can mark any logical line — answer or not (Settings & appearance).

The full reference

Everything about tokens — which answer kinds can be tokenized, the automatic previous-answer path, insertion edge cases, copy behavior and the current precision limits — is documented in Answer tokens.