Start with saved results
A notebook file is a document containing code and a snapshot of output. The fact that a cell contains Python does not mean a converter knows its dependencies, files, random state or environment. Save the results you want to distribute before importing the notebook. An empty outputs array remains empty here.
The publisher reads nbformat 4 source strings and source-line arrays. It uses one-based cell numbers for the export plan, so a duplicated or absent notebook cell ID cannot make a selected cell disappear. If your file uses an older major format, resave it in a current notebook editor first.
Choose the story and the page
Start with the explanation and the central results. The selection field accepts a range such as 1-4,7 and also preserves an explicit order such as 3,1,2. Turn off code source when the intended reader needs the findings rather than implementation details. This does not remove secrets that already appear in a saved result.
A4 and US Letter have different dimensions. Landscape can reduce the number of column panels for wide tables. Long code lines wrap instead of being clipped; their display may therefore occupy more lines than in the notebook editor. The HTML export is useful when the screen-oriented layout is more important than a fixed page.
Verify the artifact, not only the preview
Download the actual PDF and open it in your normal PDF reader. Check page count, the first and last selected cells, a wide table and at least one equation or image. Text should be selectable. Column panels identify their source columns, and row labels let you reconnect values that were split horizontally.
Keep the manifest beside the PDF and HTML. It records the selected source cells, available and chosen MIME representations, image dimensions and export notes. It does not certify that the notebook calculations are correct. Preserve the original .ipynb so a later reader can inspect or rerun the original work in an appropriate environment.