The bridge experiment
Do the basic case tests hold up in ordinary writing?
The first two episodes isolated lowercase and uppercase pangrams. This episode keeps the same efficiency question but moves it into a longer, mixed-case sample: headings, ordinary sentences, capitals, and punctuation. It is a practical continuation of the series, not a new claim about legibility or font preference.
The focused hypothesis is that more continuous joining can add paper-trail distance. The data is reported openly so readers can see where that hypothesis is supported, where pen-up travel changes the balance, and why no single metric should be treated as a universal ranking.
Measured results
One mixed-text run, eight metrics
| Metric | Getty-Dubay Joined Standard | Getty-Dubay Joined Beginning | Playwrite CA | Run result |
|---|---|---|---|---|
| Completion timeLower is faster for this run | 54.36 s | 52.41 s | 54.50 s | Getty-Dubay Joined Beginning |
| Characters per minute37 words; punctuation is reported separately | 262.69 | 272.45 | 262.00 | Getty-Dubay Joined Beginning |
| Words per minuteSame mixed-case sample for all three models | 40.8 | 42.4 | 40.7 | Getty-Dubay Joined Beginning |
| Paper-trail lengthPen-down path length | 7,322 px | 6,962 px | 7,493 px | Getty-Dubay Joined Beginning |
| Air-travel lengthPen-up travel between strokes | 3,658 px | 3,816 px | 3,163 px | Playwrite CA |
| Total movementPaper trail + air travel | 10,979 px | 10,778 px | 10,656 px | Playwrite CA |
| Words completedSame task for all three models | 37 | 37 | 37 | Tie |
| Punctuation completedPunctuation marks in the sample | 12 | 12 | 12 | Tie |
A run result names the better value for that metric only. It is not an overall font ranking.
The sample
A longer sample with real writing transitions
Handwriting Examination
The quick brown fox jumps over the lazy dog.
Always Be Careful During Exercise. Find
Good Habits. Improve Judgment. Know, Learn, Master New Opportunities
Presenting Quick Reward Sources — Things Usefully, Vibrantly Wonderful: Xylophones, Yo-yos, Zoos.
What this run shows
Continuous joining is not the same as faster writing
Getty-Dubay Joined Beginning finished first at 52.41 seconds and recorded the highest character and word throughput. Its paper trail was also shorter than the other two models. That supports the series question: a more connected path does not automatically make the task faster.
Playwrite CA recorded the shortest air travel and the shortest total movement, yet it finished slightly after Getty-Dubay Joined Beginning. In this renderer, path distance and completion time therefore describe related but distinct costs. The useful conclusion is limited and specific: joining patterns should be tested with both pen-down and pen-up movement, not assumed to be efficient from the label alone.
How the simulation works
Real stroke paths, deterministic canvas rendering
The video is an original simulation, not AI-generated handwriting. We extracted real stroke-order paths from each tested writing model and replayed those paths on canvas under matched settings.
The experiment uses the same 150 n/min setting for all three samples. Pen-up travel is modeled separately so the cost of lifting and repositioning the pen remains visible in the totals.
These are benchmark values from our renderer. They are not direct physical measurements of every writer, pen, surface, or classroom situation.
Transcript and limitations
Read the evidence without playing the video
Transcript: This mixed-text handwriting speed test compares Playwrite CA, Getty-Dubay Joined Standard, and Getty-Dubay Joined Beginning on a 37-word sample with 12 punctuation marks. All three models use the same 150 n/min setting. The test records completion time, characters per minute, words per minute, paper-trail length, air-travel length, total movement, words completed, and punctuation completed. Getty-Dubay Joined Beginning finishes fastest in this run. Playwrite CA has the shortest air travel and total movement. The animation is an original deterministic canvas simulation from extracted real stroke paths, not AI-generated handwriting.
- This is one controlled run, not a population study or a guarantee for every writer.
- The sample tests efficiency only. It does not evaluate legibility, learning outcomes, comfort, or personal preference.
- Pixel distances and the 150 n/min setting describe this renderer's benchmark model.
- No overall winner is assigned because the metrics do not share one pre-registered composite score.
Credits and permission
Original work published with permission
This independent comparison was created by Xin Wan / PrintableHandwriting with Kate Gladstone. Kate received permission from HandwritingSuccess to use the Getty-Dubay models for font research, and PrintableHandwriting received Kate's permission to publish this video, data, and explanatory page. PrintableHandwriting is not affiliated with, endorsed by, or a reseller for HandwritingSuccess.