Splay is silver streaking left by gas that reached the surface during fill. The gas comes from three places: water in an undried hygroscopic resin, polymer breaking down from heat or shear, or air and volatiles drawn into the melt. Where the streaks sit, and whether they move from shot to shot, tells you which to fix first.
Splay is cosmetic, but it is also evidence. Something turned to gas between the hopper and the cavity, the gas rode along at the flow front, and the front smeared it against cold steel, where it froze into the skin as a silvery streak. The streaks run with the flow. On a dark or glossy part they read as frosty silver; on a clear part they look like fine scratches inside the wall. Fixing it means finding where the gas came from, and the part usually gives that away if you look before changing settings.
Three defects get mixed up on first samples, and each one sends you somewhere different:
The confusing case is a streak with a brown tint. That is usually melt that is breaking down from heat, which makes gas and discoloration at the same time, so treat it as splay from degradation rather than a burn.
The troubleshooting literature lands on a short list. Garrett MacKenzie's Plastics Technology column on splay names moisture, shear and heat. The Mold-Masters guide adds trapped volatile compounds, low back pressure and screw decompression. Grouped by where the gas starts, they make three families.
Hygroscopic resins such as polycarbonate, nylon, PBT, PET and ABS hold absorbed water inside the pellet, and at melt temperature it flashes to steam. DuPont's Zytel molding guide says excess moisture in nylon forms steam that causes splay marks and internal bubbles, and that properties lost to the wet melt cannot be restored by further processing. The limits differ a lot by grade. Covestro lists a maximum of 0.02% moisture for Makrolon 2407 polycarbonate, dried 2-3 h at 120 °C. DuPont sets 0.2% for Zytel nylon, dried 2-4 h at 80 °C with dryer dew point below -18 °C, and asks for about 0.1% if melt runs above 315 °C or cycles are long. A tenfold gap between two common resins is why a dryer recipe copied from another job can be wrong.
On the part, moisture splay wanders. Tim Womer, writing in Plastics Technology, says it moves around from part to part and never sits in the same place. It can cover large areas. MacKenzie's quick floor test: purge onto a plate, and if the puddle is foamy or full of bubbles, suspect moisture. How dryers work and how to set them is covered in drying hygroscopic resins; this page is about recognizing when the dryer is the suspect.
Polymer that is overheated gives off gas as its chains break. The heat can come from barrel settings, from time spent hot in the barrel, or from shear: a fast-turning screw, a tight mixing section, a small gate or a sharp corner the melt is forced through at speed. MacKenzie points out that heat splay can look exactly like moisture splay, so look for the extra clues: a sticky part or purge, a burnt smell, a yellow or brown cast.
Shear splay has one helpful habit. It repeats in the same spot, shot after shot, and the stage of fill where it appears tells you the source. Womer gives two tests. If streaks directly in front of the gate vanish when injection speed comes down, the gate is too small for the rate. If they vanish when screw rpm comes down, the shear is happening in the screw.
Air gets into the melt when the screw pulls back too far after recovery. That suck-back, or decompression, relieves nozzle pressure, but overdone it draws air in ahead of the melt, and the first material into the cavity carries it. MacKenzie places this splay at the start of fill, near the gate, and says decompression should generally be set at 0.1 to 0.4 in. Volatiles also arrive from outside the resin: contaminated regrind, organic dirt, condensation on a mold run cold in humid air, or a weeping water line. Mold-Masters lists dirt and organics alongside moisture as contaminants, and MacKenzie lists mold water leaks.
The sketch is a starting guess, not a verdict. Two causes can stack, and a wet resin run hot shows both patterns at once.

Change one thing at a time and keep marked parts from before and after each change. Mold-Masters publishes its splay remedies as a numbered sequence that starts with drying; the order below follows the same logic: confirm the material, then the heat, then the screw, then the fill, and cut steel last.
Take a hypothetical clear polycarbonate cover. The shot, runner included, is 2.5 oz and the cycle is 40 s. It ran clean on a press rated at 7 oz of polystyrene, then moved to a 20 oz press and started showing streaks in patches with a faint yellow cast. Same resin lot, same dryer settings.
Dryer first. At 40 s the press makes 90 shots an hour. 2.5 oz is 70.9 g, so the line eats 90 x 70.9 = 6.4 kg of resin an hour. Covestro's 2-3 h drying time means 12.8 to 19.1 kg must sit in the hopper. A 30 kg hopper gives 30 / 6.4 = 4.7 h, comfortably over, and the moisture analyzer reads under the 0.02% limit. The dryer is cleared.
Then residence time. Barrels are rated in polystyrene, so convert with the melt densities in Catignani's article, 0.97 g/cm3 for general-purpose PS and 1.02 for PC. The 20 oz barrel holds 20 x 1.02 / 0.97 = 21.0 oz of PC. That is 21.0 / 2.5 = 8.4 shots queued, and 8.4 x 40 s = 336 s, about 5.6 minutes at temperature. The shot uses 2.5 / 21.0 = 11.9% of the barrel, far below Covestro's 30% floor. On the old press, 7 x 1.02 / 0.97 = 7.36 oz; 7.36 / 2.5 = 2.9 shots; 2.9 x 40 s = 118 s, about 2 minutes, with the shot using 34% of the barrel.
So the resin now spends nearly three times as long at melt temperature. The yellow cast fits. The fix is a smaller press, which is also on Mold-Masters' list. Until one is free, MacKenzie's interim step for long residence is a lower feed-zone temperature. Simple shots-in-barrel math ignores screw stroke and how the melt actually moves, so treat it as an estimate. The calculator linked above adds a correction for that.

For a part with cosmetic faces, name splay as a defect class on the drawing or quality plan and say which faces it applies to. Ask the molder to log, per lot, the dryer temperature and dew point, a moisture reading, the melt temperature and the press the job ran on. Those four records answer most splay arguments in a few lines, and they let you compare a bad lot with the approved one. Agree the limit on a reference part at first-article review. For a broader symptom-to-cause list across other defects, the defect troubleshooter ranks causes for twelve of them.
The streaks themselves are a skin-deep cosmetic defect, but the cause can weaken the part. Wet polycarbonate, nylon or polyester loses molecular weight in the barrel, and DuPont notes the lost properties cannot be restored by reprocessing. Heat-degraded melt is weaker too. If splay came from moisture or heat on a load-bearing part, test the parts, not just the look.
No. Moisture is the most common cause in hygroscopic resins, but heat, shear at the gate or screw, air pulled in by decompression, volatiles from contaminated regrind, and a leaking or sweating mold all produce it. Splay that stays in the same place on every shot is rarely moisture, because moisture splay tends to wander from part to part.
Yes, though less often. They do not absorb water into the pellet, but cold pellets moved into a warm, humid room can collect condensation on the surface. They can also show splay from overheating, shear, decompression air, contaminated regrind or a sweating mold. The diagnosis is the same: read the pattern on the part first.
Streaks that repeat just past the gate usually come from shear or from air drawn in by screw decompression, because the first melt through the gate carries both. Try a slower fill speed at the start and less decompression. If only a slower fill clears it, the gate is too restrictive and may need to be enlarged or fanned.
Treat it as a named cosmetic defect, state which faces it applies to, and agree the limit on a reference part during first-article review. Most cosmetic faces allow none. Also ask that dryer settings, a moisture reading and melt temperature are logged for each lot, so a bad lot can be traced to its cause.
Send the part file and the quantity you expect to run, and we'll review it before you commit to tooling. NDA first if you want one.
Send a part for reviewBook a 30-minute call