Is it perfectly acceptable to produce a map at a scale of 1:24,000 from a database that was digitized from a 1:100,000 scale source map?
Not at all.
When you have some data that was digitized at 1:100,000, it excluded smaller features, which were smaller than the Minimum Mappable Area.
For example, suppose you are dealing with topographic data.At 1:100,000 you would exclude large streams and trees.These features are visible at 1:24,000, and should be included in such a map.
If you just print the same data at a higher scale, then you are missing out on data, which should be visible at that scale.
Secondly, the accuracy of the data is different at these scales. Generally, you cannot make lines smaller than 0.8 mm on a printed map. This means that at 1:100,000 scale, your features are within a about 80 meters of that location. at 1:24,000, the feature should be within 20 m of the location that you have marked. Hence if you just print out your data at 1:24,000 your map would say that you are accurate to 20 m, when you are actually not.
Don’t disagree with any comments. With GIS it’s more important to define minimum mappable unit ( and minimum width of linear features) than scale per se - as scale doesn’t have the fixed/precise meaning like it used to. It’s also important to define the precision of the line work - ie how close the line work boundaries are to actual feature (relative to image although should be relative to ground). This is partly (mainly?) a function of scale the data is captured at as, as the comment above says, you can normally get linework within 0.8 mm so if the data is captured at twice the scale the precision will be half that ( assuming you take same amount of care digitising).