This is the ugly technical page. It is supposed to have too many numbers.
Pillow Science, Harder Version
A pillow is a viscoelastic support object. It has height, spring, hysteresis, heat behavior, wash behavior, and geometry. The consumer word for all of that is "comfortable," which hides too much.
Important: the current bench results are calculated prototype data. They are for site design, formulas, and ranking behavior. Product claims need retail measurements.
Hypotheses
| ID | Claim to test | Null hypothesis | Pass signal |
|---|---|---|---|
| H1 | Adjustable pillows fit more body/mattress combinations. | Adjustability leaves alignment error unchanged. | Lower side/back angle error across narrow, average, and broad shoulder models. |
| H2 | Latex and buckwheat dissipate heat better than solid memory foam. | Material class leaves thermal retention unchanged. | Lower TRI after identical heat load and cool-down. |
| H3 | Fiber pillows lose support faster after wash and compression cycling. | Wash loft loss is equal across fill types. | Lower WDI in down alternative group after wash/dry cycle. |
| H4 | Cheap pillows can win on value while losing repeatability. | Price has zero relation to repeatability or rebound. | High score per dollar but wider 95% CI and lower recovery. |
Formulas
| Metric | Formula |
|---|
Operational Definitions
| Term | Definition | Instrumentation |
|---|---|---|
| Compressed loft | Height at pillow center after 5 minutes under a 10 lb board load. | Flat board, calibrated weight, ruler/calipers. |
| Rebound | Recovered height 60 minutes after load removal divided by unloaded height. | Same measuring station, same room temperature. |
| Heat rise | Peak surface temperature increase after a 10 minute heat-pad exposure. | IR thermometer or taped surface probe. |
| Alignment error | Absolute angle error from neutral head-neck line in side and back positions. | Photo grid, plumb line, repeatable mattress surface. |
| Wash loss | Percent loft loss after label-allowed cleaning and full drying. | Before/after loft readings, drying mass check. |
Calculated Winners
Material Group Means
Calculated rows are useful for layout and hypothesis framing. Product claims need measured rows.
| Fill class | Mean PFI | Mean LDI | Mean TRI | Mean WDI | Mean 95% CI |
|---|
Hard Matrix
N = 5 calculated repeated loft readings per pillow. PFI higher is better. TRI lower is cooler. LDI and WDI higher are better.
| # | Pillow | PFI | Loft - load | Recovery | Thermal | Wash | Angles | Fit bands |
|---|
What Would Make This Real
- Buy two retail units of each pillow from normal retail channels.
- Condition every pillow for 72 hours at the same temperature and humidity.
- Run five repeated loft readings before and after compression.
- Photograph side/back alignment with a fixed camera and calibration grid.
- Publish raw CSV, pictures, dates, SKU tags, law tags, and retailer receipts.
- Retest top pillows after 30 nights and one wash/dry cycle if the label allows it.
This Gets Expensive
Raw CSV sounds noble. The bill shows up when the second pillow has to be bought and ruined in a controlled way.
Donations buy the next retail pillow, the duplicate lab pillow, and the boring supplies that make the data useful.