NIX DRONES / STORIES
THE BREADMAKER SPECIAL
NIX KITCHEN · HISTORY, SCIENCE & BREAD
The soft
loaf project.
From Japanese bakeries to my PN100: the history, the science and the pursuit of a beautifully soft slice.
Explore the story ↓BY JOHN / NIX · 29 OCTOBER 2026
Cover: AI-generated editorial illustration. Appliance appearance informed by Panasonic’s SD-PN100 product imagery. [8]
IN THIS FEATUREThe historyThe scienceThe evidenceThe methodThe recipeThe PN100The comparison bake1 : 5
FLOUR TO WATER IN MY PASTE
6%
OF MY FLOUR COOKED FIRST
250 g
TOTAL FLOUR FOR THIS LOAF
The NIX breadmaker special
Small machine. Proper ambition.
I’m quite happy for a loaf to have a personality. Bombay mix, raisins and honey certainly give this one enough to be getting on with. But underneath the sweet and spicy mischief, I want something beautifully simple: a soft white slice that welcomes a generous spread of butter.
That is the next job for my Panasonic PN100.
The flavour idea is already there. What interests me now is the bread around it. A little cooked flour seems an almost suspiciously modest answer to the demand for a softer loaf. So I’ve followed the story through a publisher’s catalogue, Japanese breadmaking, early food blogs and the experiments behind the kitchen advice.
My rules stay familiar. A maximum of 250 g flour in total. Liquids first, as I prefer. A small loaf with a clear plan. The actual bake will have the final word. Preferably through a thick slice.
Start here
A small paste with a useful job.
Tangzhong cooks a small portion of a bread recipe’s flour and liquid into a paste, changing how the starch handles water. The cooled paste joins the main dough. A common flour-to-liquid ratio is 1:5 by weight. [2]
It is sometimes called a starter, but this paste has no separate fermentation to feed or maintain. The yeast in the main dough still supplies the gas. The saucepan prepares an ingredient; the rest of the breadmaking process still has its work to do. [1]
For my loaf, that means a saucepan holding just 15 g flour and 75 g water. Before getting to the weighing, though, there is a much better story here than “add a clever paste and hope”.
01 / The documented history
A loaf with a longer backstory.
The modern story has several strands: Japanese commercial breadmaking, a Chinese-language cookbook, and home bakers passing recipes around online. These dates document parts of that story. They do not establish a single inventor or a continuous chain linking every recipe.
Pasco brings Chojuku to the supermarket shelf.
Pasco says its Chojuku loaf launched in 1998 after the company adapted an existing yudane method for large-scale production. The problem was consistent quality: scalding some flour was familiar, but turning that into a reliable factory process took further development. Pasco’s account describes more than six months spent achieving mass production. That is a specific industrial milestone, rather than evidence that the company invented scalded flour. [10]
The cookbook date we can verify.
Wan Li’s publisher catalogue lists 65℃湯種麵包 by 陳郁芬 Yvonne — Yvonne Chen in 2004. It provides an ISBN and a publication record, which is firmer ground than an unsourced origin story. This establishes the date of that edition; it does not, by itself, settle the first publication date of every edition worldwide. The catalogue leaves its English-title field blank. [9]
The English-language trail is already visible.
On 3 August 2007, SeaDragon’s Café of the East published a Japanese-style bun dough explicitly adapted from Chen’s book. The author described recipes already circulating in Chinese-language baking communities. This surviving post gives us an early, dated example of the method crossing into English-language home baking. It is evidence of dissemination, without claiming to be the first English mention ever. [11]
A food-blog audience grows around it.
Christine Ho’s bacon-and-cheese bread post, dated 2 March 2010, also credits Chen’s book and describes a tangzhong dough mixed in a breadmaker. Its English shorthand for the book includes “Bread Doctor”. For this history, the useful evidence is the dated recipe and attribution: another visible point in the method’s online journey. [12]
A wider baking introduction.
King Arthur Baking published PJ Hamel’s introduction to tangzhong on 26 March 2018, explaining it to an English-speaking baking audience. By then, the earlier book and blog trail were already established. The date marks another stage in its popularisation. [1]
That leaves a pleasingly human history: bakers adapting a technique, a book giving it a recognisable identity, and home cooks sharing what they made. I can trace those steps with confidence while leaving the unproven invention story open.
Two related techniques
Tangzhong and yudane: close relatives.
In contemporary English-language recipes, these names commonly distinguish two preparations. Both bring heat and water to flour before the main dough is mixed. [6]
| Tangzhong | Yudane |
|---|---|
| Flour and liquid cooked together in a saucepan. | Boiling water poured onto flour and mixed. |
| Commonly 1 part flour to 5 parts liquid by weight. | Commonly 1 part flour to 1 part water by weight. |
Recipes vary in ratio, flour fraction and resting time; an overnight rest is not a universal rule for every version. Equal weights of these two preparations contain different amounts of flour and water. I would follow a recipe written for the chosen method, or recalculate both totals before switching.
02 / Inside the saucepan
The starch is where the story begins.
Flour brings starch as well as protein to the dough. Starch contains two glucose-based polymers: amylose, comparatively linear with limited branching, and amylopectin, a much larger, highly branched molecule. Their arrangement gives native starch granules regions of order. Heat and available water can disrupt that arrangement. [14]
A / BEFORE HEATING
A dispersed mixture.
The flour is suspended in water. The starch granules still contain ordered regions.
B / DURING HEATING
Structure changes.
Granules swell as water enters. Molecular order is disrupted and some starch chains leave the granules.
C / THE COOKED PASTE
A thicker mixture.
Swollen granules and dispersed starch polymers contribute to the paste’s changed consistency.
Simplified schematics, not microscope images. Orange shapes represent granule regions, graphite lines suggest starch chains, and blue dots suggest water. Sizes and quantities are illustrative; no fixed water-uptake multiplier is implied. [14] [15]
This heat-and-water-driven loss of starch order is gelatinisation. It involves swelling and the movement of some starch molecules out of the granules. The familiar kitchen result is the transition from a thin slurry to a paste. [15] Cooling does not simply rebuild the original granules. [14]
Why cook a small portion before mixing?
The cooked portion changes dough handling before the baking stage. A suitable formula can accommodate more liquid while remaining manageable, but that liquid must be planned into the recipe. [2] For me, this makes water accounting part of the method from the start.
For my PN100, the useful question is practical: can this small cooked portion help me arrive at a tender crumb without turning the rest of the dough into something the machine struggles to mix? The saucepan is one part of that answer. The flour, total water and full programme are the rest.
A number worth understanding
65°C is a signpost.
The number appears in Chen’s book title, and many practical tangzhong guides use around 65°C as a reference point. In the kitchen, the visible change matters too: the mixture should become a smooth paste that holds the trail of a spoon or whisk. [3]
Gelatinisation is a process, rather than an instantaneous change shared by every granule. In a 2016 wheat-starch study, Wang and colleagues found that ordered structures could remain even after the endpoint of a measured thermal transition. A temperature reading alone therefore cannot establish that every part of the starch has completely changed. Water availability and heating conditions belong in the explanation. [15]
I’ll use gentle heat, stir right across the base and stop when the paste is properly thickened. A thermometer can help me observe what is happening. I won’t keep boiling the mixture to pursue a supposedly perfect molecular finishing line, or keep reducing it until the water calculation no longer means much.
03 / From paste to loaf
Softness still needs a structure.
A slice can feel soft because it takes little force to compress, yet have a fairly close crumb. A lofty loaf can have a resilient or chewy bite. I want tenderness and an airy structure together, so I need to pay attention to gas retention as well as moisture.
The research makes that distinction useful. Yamada and colleagues’ 2015 model experiment combined gelatinised wheat starch with separately prepared gluten. It produced lower gas retention and lower loaf volume relative to mass; the authors linked this to interference with gluten-network formation. [16]
Their complementary 2017 experiment heated the gluten component and combined it with unheated starch. Under those conditions, gas retention and specific loaf volume increased. These deliberately simplified experiments help separate ingredients’ effects. They also show why a slogan such as “heating flour destroys the gluten, so the loaf must be smaller” is too crude. [17]
04 / What happens tomorrow
Bread can firm while the water is still there.
Drying matters, but it is only part of bread staling. As cooked starch cools and bread ages, starch chains can associate again. That reorganisation is called retrogradation. Amylopectin changes are especially relevant to crumb firming during storage. [18]
Bread experiments by Besbes, Le Bail and Seetharaman connected storage with changes in firmness, amylopectin retrogradation and water mobility. They also found that the heating experienced by different parts of a loaf matters. Texture emerges from several interacting changes, rather than a single meter labelled “water left in bread”. [18]
Baik and Chinachoti made the moisture question especially clear by storing white bread with and without its crust in sealed pouches. Keeping the crust changed the way moisture redistributed from the crumb and contributed to later texture differences. Water can move within bread even when the loaf is enclosed. [19]
That helps explain why I want observations at two moments: once the loaf is fully cool, and again the following day under the same storage conditions. A soft first slice and good texture tomorrow are related ambitions, but they deserve separate notes. I’ll judge eating quality; softness alone cannot tell me how long a loaf is safe to keep.
A small but interesting side story
Where the sweetness comes in.
Wheat flour contains enzymes that can act on accessible starch. In a 2021 study, Shintani and colleagues followed maltose formation in flour-and-water mixtures held warm under controlled conditions. They measured maltose changing over time and explained the role of the flour’s own amylase enzymes. Maltose can contribute sweetness and provide a substrate for yeast. [20]
That is more specific than saying heat simply turns the paste into sugar. The study’s warm-holding protocol is different from my brief saucepan step, so I cannot calculate a sweetness bonus for this loaf from its results. My honey and raisins already have that part of the brief well covered. [20]
05 / Put the claims beside the measurements
What the experiments actually found.
Parenti and colleagues (2019) tested salt-free brown bread in a Moulinex breadmaker: 6% cooked flour, a 1:4 paste heated to 85°C for three minutes, then chilled overnight. The matched-consistency formulas used 61% (control) and 64.5% (treated) water relative to flour. [21]
Means ± standard error; Parenti et al. (2019), Table 4. Published measurements, not NIX results. [21]
The treated formula gave greater volume and lower hardness, with different water levels. Hardness time and treatment-by-time effects were not significant, so these observations cannot establish slower firming or predict my PN100 result. [21]
Other recipes give a different picture.
Yamauchi and colleagues’ 2014 yudane study found softer bread and slower firming, alongside lower specific loaf volume. Its formula and preparation differed from the experiment above. Taken together, the results give a good reason to keep “softer” and “taller” as separate observations. [5]
There are useful kitchen counterexamples too. In 2019, King Arthur’s PJ Hamel reported no texture improvement from tangzhong in three sourdough recipes she tried. That is a practical trial, rather than a universal verdict on sourdough. It is enough to remind me that an extra step needs to earn its place in the loaf I actually bake. [22]
06 / Back to my small machine
The flour comes out of the total.
A common starting point is to cook 5–10% of a recipe’s flour with five times its weight in liquid. These are useful working proportions, with variations between recipes. [2] For this NIX loaf, I’m choosing 15 g flour and 75 g water.
250 g flour, counted once
15 g in the paste · 6%
235 g in the main dough · 94%
185 g measured water, counted once
75 g in the paste
110 g added to the pan
Bars show the ingredient split in this proposed NIX formula. They do not measure water binding or predict loaf softness.
The flour total stays at 250 g. I weigh that first, then take the 15 g paste portion from the bowl. The measured water totals 185 g, so the water-to-flour ratio is 74%. Honey and butter add some further moisture; this percentage describes the weighed water, not every drop contributed by every ingredient.
Recipe conversion can require more water to keep the dough workable. King Arthur’s roughly 75% starting guidance for suitable soft yeast breads is not a universal setting. [2] In its sourdough trial, 67% worked better than 75% for one tested dough. [22]
Water keeps this first version straightforward. Tangzhong can also be made with milk in a recipe formulated for it. [3] I’ll keep this batch’s ingredients fixed long enough to learn something, rather than change the liquid, flour fraction and add-ins all at once.
The method / before loading the breadmaker
One small saucepan. Four simple steps.
For this 250 g flour example: 15 g strong white bread flour + 75 g water.
Both quantities are already included in the loaf’s totals. You also need a small saucepan, a whisk or spatula, a bowl and scales.
Mix while cold.
Whisk the 15 g flour and 75 g water in the saucepan until smooth. Starting cold makes it easier to disperse the flour evenly.
Heat gently and keep stirring.
Stir over a gentle heat, reaching the base and corners. Stop when the mixture becomes a thick, smooth paste and the spoon or whisk leaves tracks. Avoid cooking it dry. [4]
Cool before it meets the dough.
Scrape the paste into a bowl and cover its surface to limit drying. Let it cool to room temperature before loading the bread pan. [4]
Keep the water balance repeatable.
The starting mixture weighs 90 g. For repeatability, I’ll replace any water lost to evaporation: scrape the pan thoroughly, weigh the cooled paste without its bowl, and stir in enough water to return it to 90 g. Use the whole paste. This corrects evaporation, not flour left stuck in the pan. [6]
Many guides mention around 65°C. Visible thickening is the useful kitchen cue; a thermometer is optional. [3]
Worked example / the paste is included
Bombay Honey Riot: the soft-loaf formula.
One small PN100 loaf · 250 g flour in total.
Development formula for the first test bake.
| Item | Paste | Pan | Total |
|---|---|---|---|
| Flour | 15 g | 235 g | 250 g |
| Water | 75 g | 110 g | 185 g |
For water, those amounts are approximately 75 ml and 110 ml. Weighing them keeps the 1:5 paste ratio easy to follow. The 185 g total belongs to this formula.
The rest of the dough
| Ingredient | Amount |
|---|---|
| Honey | 25 g |
| Unsalted butter, softened | 20 g |
| Skimmed milk powder | 10 g |
| Fine salt | 4 g |
| Instant dried yeast | 1 level teaspoon |
| Bombay mix, lightly crushed | 30 g · at the addition beep |
| Raisins | 20 g · at the addition beep |
Menu 3 · Basic Raisin · Light crust
Full programme: about 4 hours–4 hours 10 minutes, plus making the paste and cooling the loaf. [7]
Keep the flour and water in the paste within the totals above. The machine handover below takes the recipe from here.
The NIX breadmaker routine
From saucepan to PN100.
This is where the extra step earns its place in the routine. Make the paste, let it cool, then organise the pan. King Arthur’s milk-bread recipe includes bread-machine mixing, so the technique can fit a machine-assisted loaf. [4]
Load it, liquids first.
My preference is the 110 g water and honey first, then all the cooled paste, butter and salt. Add the remaining 235 g flour and milk powder, with the yeast dry on top until mixing. Panasonic’s published order puts yeast in first; this is my NIX loading preference. [7]
Let the quiet beginning happen.
Choose Menu 3 — Basic Raisin, with a light crust, then press Start straight away. The normal initial rest belongs to the programme. It does not need me hovering over the lid like a suspicious harbourmaster. Let the full cycle run. [7]
Look during active kneading.
For the first recipe-development bake, after several minutes of active kneading, I’m looking for a soft, cohesive dough with a little tackiness. If it remains distinctly stiff and ragged, I’ll add 5 ml water, let it mix in and reassess. I’ll note any extra water. My flour total stays at 250 g. A controlled comparison bake needs the stricter water rule described below.
Add the personality at the signal.
In go the prepared 30 g Bombay mix and 20 g raisins. Keep hard pieces small and chop large raisins. Close the lid and press Start to resume. When the programme finishes, remove the loaf promptly and let it cool on a rack. [7]
My four reminders: whisk smooth, stop before the paste dries, cool before loading, and count the paste’s ingredients once.
07 / The useful notebook questions
Keep the first bake interpretable.
The paste is lumpy or has dried out.
I’ll aim for a smooth cold mixture, then stir while heating. If the result is scorched or full of dry lumps, remaking a 15 g flour batch is a clearer starting point than building the whole loaf around it. A smooth paste and thorough transfer make the written formula meaningful. [4]
The finished loaf is soft but quite close-grained.
I’ll record that as two observations: tender crumb and modest aeration. I’ll check the ingredient weights, yeast, programme and what the dough looked like during kneading before deciding that more paste is the answer. The published studies already show that softness and volume need not move together. [5]
The mix seems wetter or stiffer than expected.
For recipe development, small measured adjustments can help me find a workable formula. I’ll write each change down and use the revised total next time. For the separate comparison test, an unrecorded splash would make the result harder to explain.
The photograph looks smoother than my Bombay mix loaf.
The images show the texture ambition. My actual loaf contains 50 g of additions and will have its own crumb and appearance. I’ll keep the same Bombay mix and raisins across comparison bakes, so a change of brand or a generous handful does not become another moving part.
Should I cool the loaf before judging it?
Yes. I’ll let it cool on a rack, then compare slices of the same thickness at the same age after baking. The comparison should answer how the bread eats as a slice, including the following day. I’ll seal it only once it is fully cool.
08 / The next NIX experiment
Give the saucepan a fair test.
First I want a workable loaf. Then I can ask a narrower question: what changes when I cook that flour-and-water portion first? Those are two different jobs. Optimising a recipe allows adjustments; a comparison needs a clearly held baseline.
Loaf A / with tangzhong
Cook the small portion.
Make the 15 g flour + 75 g water paste. Restore its cooled weight to 90 g after thorough transfer. Add the remaining 235 g flour and 110 g water, plus the other ingredients in the recipe.
Loaf B / without tangzhong
Use the same totals.
Load 250 g flour and 185 g water without cooking a portion. Use the same honey, butter, milk powder, salt, yeast, Bombay mix and raisins as Loaf A. Keep the programme and crust setting the same.
I’ll use the same flour bag and ingredient brands, start with the pan and ingredients at comparable temperatures, and note the room conditions. With one PN100, the bakes will be sequential. I’ll compare them at the same age after baking, then repeat in reverse order if the difference looks worth pursuing.
For this same-water comparison, I won’t quietly correct one dough and leave the other unchanged. If a bake needs extra water to become workable, I’ll log it as a development bake and design the next comparison around an explicit question. Matching dough consistency by changing water tests the usefulness of the adjusted recipe, rather than heating alone.
The record can stay simple: final water total, a kneading note, cooled loaf weight and height, a photograph of the middle slice, then tenderness and flavour notes after cooling and after 24 hours. Same slice thickness, same storage. A hand squeeze is a personal observation, not an instrument reading in newtons.
If I later want to separate heating from prehydration, a third loaf could use the same 15 g flour and 75 g water mixed cold for the same preparation interval. For now, two carefully recorded loaves are a useful first comparison. The result may be a clear benefit, a subtle preference or no worthwhile difference; each would tell me what to do next.
The NIX verdict / still to be baked
The last word belongs to the slice.
I started with a slightly unruly flavour idea and a straightforward wish: make the white bread around it wonderfully soft. Tangzhong has given that wish a proper investigation, from a documented cookbook and bakery history to starch science and useful experiments.
Now I have a method to try, a formula I can account for and a comparison worth making. What I don’t yet have is a tested PN100 result. That comes after the machine has finished and the loaf has cooled.
The Bombay mix brings the personality.
I’m giving the bread the same attention.
If the crumb is tender, the slice holds together and the sweet-spicy idea still makes me reach for another piece, that will be a good day in the NIX kitchen. The butter dish remains on standby.
Research desk
Sources & further reading.
The history uses dated publisher, manufacturer and original blog records. The science draws on original research, with practical baking guides kept distinct from laboratory findings. The 250 g flour ceiling, recipe quantities and proposed comparison are NIX development choices.
The history
- 65℃湯種麵包 — publisher catalogue record
Wan Li. Author: 陳郁芬 Yvonne; edition dated 2004; ISBN 9789621428585. Primary bibliographic evidence. - What is the Chojuku method?
Pasco. Its own account of adapting yudane for the Chojuku bread launched in 1998. - Japanese-Style Bun Dough 湯種麵糰
SeaDragon · Café of the East · 3 August 2007. An early English-language recipe explicitly adapted from Chen’s book. - Japanese Style Bacon and Cheese Bread
Christine Ho · 2 March 2010. Evidence of the book and method’s continuing online dissemination. - Salinātā rudzu rupjmaize — traditional-speciality application
Official Journal of the European Union · 22 June 2013. Latvian scalded rye-bread tradition, including a 1931 recipe reference. - Tangzhong, the secret to softer bread
PJ Hamel · King Arthur Baking · 26 March 2018. An identifiable milestone in English-language baking coverage.
Starch, texture and the published experiments
- Starch structure and physicochemical behaviour — doctoral research abstract
Greet Vandeputte · KU Leuven · 2004. General starch definitions; the experimental material was rice starch. - Changes of multi-scale structure during mimicked DSC heating reveal the nature of starch gelatinization
Wang et al. · Scientific Reports 6, 28271 · 2016. DOI: 10.1038/srep28271. Wheat-starch structural measurements. - Effect of Gelatinized Wheat Starch on Bread-making Qualities of Yudane Dough
Yamada et al. · 2015. DOI: 10.3136/nskkk.62.547. A model experiment separating the starch contribution. - Effect of Heated Gluten on Bread-making Qualities of Yudane Dough
Yamada et al. · 2017. DOI: 10.3136/nskkk.64.90. A complementary model separating the protein contribution. - Impact of local hydrothermal treatment during bread baking on soluble amylose, firmness, amylopectin retrogradation and water mobility during bread staling
Besbes, Le Bail & Seetharaman · 2016, online 2015. DOI: 10.1007/s13197-015-1992-z. - Moisture Redistribution and Phase Transitions During Bread Staling
Baik & Chinachoti · Cereal Chemistry 77, 484–488 · 2000. DOI: 10.1094/CCHEM.2000.77.4.484. - Maltose formation from wheat flour with different degrees of damaged starch
Shintani, Miyagawa & Adachi · 2021. DOI: 10.3136/fstr.27.567. Enzyme activity and starch condition during controlled warm holding. - Cooked flour and brown wheat bread quality
Parenti et al. · LWT 106, 240–246 · 2019. DOI: 10.1016/j.lwt.2019.02.066. Charts: T2, Table 4. - The Staling and Texture of Bread Made Using the Yudane Dough Method
Yamauchi et al. · Food Science and Technology Research 20, 1071–1078 · 2014. DOI: 10.3136/fstr.20.1071.
The kitchen and the machine
- How to convert a bread recipe to tangzhong
PJ Hamel · King Arthur Baking · 2018. Practical flour fractions, paste ratios and hydration guidance. - A closer look at tangzhong
Rossi Anastopoulo · King Arthur Baking · 2021. Common methods, liquid choices and the familiar temperature cue. - Japanese Milk Bread
King Arthur Baking. Paste consistency, cooling and a bread-machine mixing option. - Tangzhong & Yudane Explained, Bread Improvers
ChainBaker. Practical methods, variations and accounting for evaporated water. - Tangzhong in sourdough bread
PJ Hamel · King Arthur Baking · 8 January 2019. A practical three-recipe trial reporting no improvement. - SD-PN100 operating instructions and recipes
Panasonic. Published loading order, programme timing, rest and the addition signal. - SD-PN100 compact breadmaker
Panasonic UK. Appliance appearance reference for the generated cover.
Research checked 29 October 2026. Links lead to the source documents and pages; they do not validate this particular PN100 recipe.
NIX BREADMAKER SPECIAL / THE SOFT LOAF PROJECT
nixdrones.com ↗