Covid-19 and its overall effect on England and Wales death rate.

Covid-19 has affected death rates in obvious ways: it has been a direct cause of death; it has occurred alongside other illnesses, resulting in some earlier deaths as a consequence; its disruptive presence throughout the health service has resulted in early deaths for some who have been denied life-saving treatment, or have been reluctant to seek treatment, for other conditions. As 2020 comes to a close, we can see the overall effect of covid over getting on for a whole year and its overall impact on the death rate for England and Wales.

This graph shows the erratic nature of the weekly death figures. There are big spikes down at Christmas (and other holidays) when deaths are reported late and up just after New Year when the deaths over Christmas are added in. But this graph already shows some interesting figures.

The death rate for over 65s is about 4%. That means that 1 in 25 over-65s die within a year on average. But since there are about 11 million people over 65, that is about 440,000 over-65s deaths per year.

The death rate for under-65s is about 0.2%. We can see that that it has varied little as a result of covid. On average 1 in 500 under 65s die every year. And since there are nearly 50 million under-65s, that means that there are about 100,000 under-65s deaths per year.

The average total death rate for England and Wales is the 440,000 over-65s plus the 100,000 under-65s, giving a total annual death rate of about 540,000.

If we average the weekly death rate over a 4-week period, we see more of the underlying pattern.

This shows that the death rate for over 65s varies quite a lot over the year, hitting a peak sometimes over 6% in the winter, dropping to below 4% in summer. This year we had a very high peak of over 8% in March but that was followed by particularly low death rate in August. Some of those who died in the April covid peak did not live to die at their expected time of August. This provokes the question as to what extent the reduced deaths of August compensated for the increased deaths in May. To find out the extent of this balance, we average over a longer period of 4 months.

The 4-month average shows that 2020 is only a slightly unusual year. For the over-65s, the death rate rose to a high of 5.6%, while falling to an unusual low of 3.6% in September.

Certainly covid was instrumental in the particularly high death rate in April but there was another contributory factor: winter 2018-19 was a year in which there were relatively few flu deaths. This means that the vulnerable who would normally have died in that winter were still around in the peak of the 2019-20 season and among those who succumbed to covid.

We still see the particularly low death rate of 3.6% in September, a consequence of the vulnerable dying in April and not being around to contribute to the September statistics.

Monthly figures are not available for years before 2010. But whole year figures are available from 2006 and are shown here.

As can be seen, the 2020 death rate is entirely normal in the historical context, having an overall death rate for over 65s of 4.5%, slightly more than the peak of 4.4% in 2015 but less than the 4.8% figures from 2006 to 2008

Calculating total excess deaths
On the basis of previous years we predict how many deaths we would expect in a given year and compare it with the actual number of deaths. These figures we have been dealing with enable us to make a prediction for death rates for 2020.

If we confine our calculations to the recent low death-rate years, the average death rates for under-65s from 2013 week 52 to 2019 week 51 are as follows:
Under-65s – 0.169%
Over-65s – 4.321%

Using the figures below for the numbers in the two cohorts, this gives predicted deaths for 2020 of
Under-65s – 83,376
Over-65s – 459,545
Total – 562,922 (The extra 1 being a consequence of rounding issues.)

In fact the total deaths in England and Wales for 52 weeks from 2019 week 52 to 2020 week 51 has been 600,058. This is an excess death figure of 37,136. However, if we compare 2020 with 2006-8, we find that 2020 has had 27,491 fewer deaths pro rata than these years.

Even if the excess mortality in 2020 is close to 37,000, that includes deaths from all causes, including the knock-on excess deaths from all other diseases where treatment has been disrupted by attention to covid. (Many of us know perhaps more individuals who have had life-saving treatment disrupted than individuals who have died of covid.)

Bearing in mind that there have been suggestions that excess deaths for cancer may be, at a minimum, in the order of 10,000, it seems likely that the excess deaths due to covid will be well under 30,000. This is in stark contrast to the figure published for covid deaths in England and Wales of 65,795 (as on 31/12/2020).

As can be seen from the graph below about excess winter deaths, seasonal deaths commonly vary by over 20,000 from one year to another. In which case, even were the 37,000 excess deaths this year all due to covid, that would not be far out from the range of swings that habitually occur from year to year – see the graph below.

Survivability – the chance of surviving the year
So far we have calculated in terms of death rates, the chance of dying in a year. But a different, and possibly more relevant, perspective is to consider the chance of surviving the year, a more useful way of seeing the impact of covid on our lives.

For those of us over 65, over the 6 years from 2013 to 2019, the death rate was 4.3%, meaning that our chance of survival was 95.7%. 2019 was a particularly good year in which survivability of over-65s was 95.9%. In 2020, with a death rate of 4.5%, survival rate has been 95.5%.

Here we can see how survivability has changed over the last fifteen years.

With over-65s survival rate consistently a little above 95% we can see that the effect of covid has been really very small. We can stop worrying and regard 2020 as a normal year.

Since first publishing this, the Office for National Statistics has crunched the figures. Their figures are for the 12 months to the end of November 2020. Here is the graph for age-standardised-mortality.

As you can see, this graph is pretty much identical in shape to my own graph on death rates above. Again it shows that 2020 is not a particularly exceptional year.

The BBC, in their publication of the ONS statistics, were very naughty. They focused on the excess deaths in 2020 being greater than any other year since the Second World War. Well, since the population (at 68 million compared with 47 million during WW2) and the proportion of over-65s (18% as opposed to 10%) are greater in 2020 in any other year, of course we should expect the death rates to be greater.

Statistical notes
These figures are produced by taking the ONS published weekly death rate figures, multiplying them by 52 to give an annual death rate figure and dividing them by the numbers in the two cohorts (0-64 and 65+).

Numbers in the two cohorts are calculated from published UK population figures, reduced by 11.3% decreasing to 11.0% from 2010 to 2020 to account for those in Scotland and Northern Ireland.

Percentage of over-65s in the UK population is derived from the sources below, interpolating between fixed point data where necessary. Here are the figures used.,from%202001%20with%208.3%20million.

Multiple bank accounts

For a long time I was a believer in a single, well-checked bank account. I realised now that I’d never checked most of my spending. I had a rough idea how much it was because I noticed whenever I took out £100 or so from the bank and how long it had been since I last withdrew cash. At any instant I could see how much cash I’d spent by how much was left in my wallet. But I never checked small amounts like the 49p for cable clips, or the £1.50 for a pasty, because I never checked individual receipts for cash payments.

I always had a rough idea how much cash I spent which I checked by matching 4 or 5 cash withdrawal slips with the bank statement, at the same time checking the dozen or so big items which were not cash, and my total cash spending from the cash withdrawal

Then the world changed as the number of card payments from my ‘single, well-checked bank account’ steadily rose. Come bank statement checking time, I found it was a nightmare with so many bits of paper. The big items were being disguised by a large number of small items, like 49p for some cable clips or £1.50 for a pasty. Bank account checking, which had been easy, became difficult. That coincided with a suggestion from a family member that multiple bank accounts were worth considering and eventually I was converted. I now have four main bank accounts:


Income account
All my income goes into my Income account. From that account there are three standing orders to my other three accounts. It takes seconds to check my Income account statement because just my income goes in and three standing orders come out.

Bills account
I have a Bills spreadsheet (Google docs) which lists all my regular bills, including estimated amounts for things that I know will happen. It’s something like this:

The spreadsheet tells me that my regular bills are something around £650 a month and there is a standing order from my Income account to my Bills account of £700 a month. It’s easy to check my Bills account because there is the single income from my Income account and a number of payments, all of which I am expecting and included on the spreadsheet. As the years go on, I add more and more things to the Bills spreadsheet which gradually gets clearer as a predictor of the regular outgoings.

Savings account
There is a standing order from my Income account straight into my Savings account, of an amount I have decided to save every month. (In fact there is another standing order into the Savings account of £150, the car replacement cost and the family holiday cost. Those amounts, too, accumulate in my savings account and, come new car time and holiday time, those bills are paid from my savings account.) Checking my savings account is easy. There are the two standing orders, one from Income and one from Bills, and that’s it until I spend on holiday or car.

The third payment from my Income account is to the Day-to-day account. From that I pay food, car fuel, meals out, any hobby expenditure, paint for the house, furniture, books, magazines, Amazon expenditure, Ebay, optician charges…plus cable clips and pasties. In other words, all the other expenditure. My credit card comes out of this account as well. So all my expenditure is summarised by the Bills account or the Day-to-day account, through which all my expenditure is paid.
My day-to-day account is a Starling account, though Monzo is just as good. Starling is an electronic banking account that is very easy to use and check. It is my wallet and gets filled up by the standing order at the beginning of the month. At any stage of the month I can look at it and see how much I have spent and how much I have left.
Do I check the Day-to-day account? No I don’t. It’s like I used to treat my wallet. I only look at the totals. But it’s better than my wallet because I can look back at the expenditure over the month and instantly see the patterns in my expenditure. Starling provides me with monthly summaries categorising my expenditure as well as the detail if I want it.

The consequence of all the above is that I check Income, Bills and Saving monthly. But it only takes seconds, because they are so easy to check. I check Day-to-day quite often but briefly, seeing how long it is until the end of the month and how much I have left. As for the nightmare of monthly checking of my bank statements, that has gone completely. Life is so much easier and clearer.
Is this OCD? Possibly. But actually it is appropriate OCD. Appropriate OCD makes life easier, reducing the time doing things that one doesn’t want to do, leaving more time for the things that one does want to do.

Other accounts
For a while, bank accounts have been free. So, if a different type of expenditure arises, it’s been easy to have another account for that expenditure. Even where there is a charge (Starling charges £2 a month for extra accounts) it can be worth having additional accounts if you can’t get a free one from a different bank. House renovations: get a house renovation account. Pay for everything with a card from that account and instantly you can see how much has been spent on the whole project. If it’s easier, and to avoid confusion, it’s possible to have separate car and holiday accounts to keep them separate from general savings. Husband and wife ‘pocket money’, separate from general household expenditure, can also go in separate accounts.
Other forms of income – share dividends, self-employed work etc – can also each have their own account. That way, come tax return time, all the information is in one account.

Power tools – battery vs mains

Here is a very rough comparison of the costs of mains and 18V power tools. The comparisons are between Einhell, Makita, deWalt, Milwaukee and Bosch. The comparisons are rough and ready, using the cheapest available items in each category in Screwfix, Toolstation and Wickes online as sources for the prices. There was no attempt to match specifications, for instance by ensuring that all were brushless, etc. Milwaukee corded tools are not included because they do not seem to be generally.

Don’t compare directly the Total, excluding battery, column for corded and cordless, because the former does not have an impact driver included.
The brands are in price order with Einhell by far the cheapest and Bosch most expensive.

1. There is not an awful lot of difference between the main brands, particularly since I noticed that the Bosch cordless tools are particularly highly specified, which accounts for some of the price premium over the others. A friend whose judgement I trust says that the Milwaukee quality is better than Makita.
2. The cordless premium is calculated by subtracting the corded prices in the corded column from those in the cordless. As can be seen, the cordless premium is relatively low (around 10% when one takes into account the previously mentioned high specifications of Bosch cordless). So, once one has the batteries and charger, one might as well go cordless.
3. All brands have one or two cheap offers which bundle a couple of batteries and a charger, making the transition to cordless less than the prices above indicate.
4. Makita LXT was launched in 2005 and is still their major brand. So the technology is mature and obsolescence of currently purchased tools unlikely to be a problem.
5. Makita is a market leader and it may be that its ubiquity stimulates competition which tends to drive the price down.

Relevant reviews

On Trustpilot, Einhell seems to get poor reviews, whereas Einhell UK gets good ones. Bizarre. Toolstation has variable reviews of Einhell. Wickes has generally good reviews of the brand.

Makita seems to have chuck problems. See for instance:

Other random thoughts
Mainstream manufacturers are also producing 10.8 V or 12 V tools in addition to 18 V. I suppose that ‘small and light’ must be the aim. At the same time there is a tendency to move to higher voltages for beefier tools, sometimes double 18 V and sometimes treble. So the battery market seems to be getting more diverse rather than more standardised.

Container ships…

…and the business of transporting manufactured goods across the world.

Seeing this video of the container ship Maersk Essex, I couldn’t resist trying to count how many containers she seemed to be carrying.

At the rear we see the stack is 19 containers wide and probably 10 containers high (guessing on how deep they go into the body of the ship) and it seems that the ship is about 20 containers long. That’s 19 x 10 x 20 = 3800 containers altogether. These containers are 40 ft long, twice the standard unit for calculating container loads, which is the TEU (Twenty-foot Equivalent Unit), so our calculation would seem to show that the capacity of the ship is 7600 TEU.

In fact our calculation is a significant underestimate. As this photo shows, containers are stacked well inside the hull of container ships.

The website gives more details of the MAERSK ESSEX. She was was built in 2011 and is sailing under the flag of Denmark. Her carrying capacity is 13,100 TEU and her current draught is reported to be 12.5 meters. Her length overall (LOA) is 366.44 meters and her width is 48.26 meters. At the time that I am writing, 24 November 2020, says she is on her way from Los Angeles and due to arrive at Yokohama on 2 December.

Consider the Maersk Essex to be fully loaded. That’s 366 x 48 x 12.5 m³ = 220,000 cubic metres and since each cubic metre of water has a mass of 1 metric tonne (1t), the total mass of water displaced is 220,000 t.

In fact the ship is not uniform and rectangular so in this case we have overestimated the displacement. In fact her Gross Tonnage is 141,000t.

Panamax and New Panamax ships

The Maersk Essex is an interesting size. She is just within the size for the new, wider, deeper Panama canal, which opened in 2016. The new canal sections increased the size of ships from 5000 TEU to 13,000 TEU, these sizes of ship being referred to as Panamax and New Panamax respectively.

There are significantly larger container ships. The HMM Algeciras built in 2020 is 400 m long and 61 m wide. It has a gross tonnage of 228,283t and a container capacity of 23,964 TEU. Ships like this which are larger than will fit through the Panama Canal are called post-Panamax or super-Panamax.

Suez canal

The Suez Canal now allows ships up to 400 m long with a beam of 50 m and a draft of 20.1 m. This is larger than New Panamax, leading to a maximum ship size of 160,000 tonnes, and a container load of around 14,500 TEU. In fact the Suez Canal has no locks and so there is no engineering limit on ship length, despite the regulation limit of 400 m.

Other pinch points around the world

Malaccamax, about 300,000 tonnes is a term used for the largest ships that can get through the shallow 25 m deep Strait of Malacca between Sumatra (of Indonesia) and Malaysia. Seawaymax, only 28,500 tonnes, describes the largest ships that can travel through the canal locks of the St Lawrence Seaway.

Standard container sizes

What we often refer to as a Shipping Container is technically an ISO Intermodal Container, designed for transporting goods in ships, lorries and trains without the goods being individually loaded or unloaded. The ‘standard’ ISO size is 20 feet (6.1 m) long x 8 feet (2.43 m) wide x 8 feet 6 inches (2.59 m) high. This is technically 1 TEU. Most containers are twice as long as that and a 40 ft container is obviously 2 TEU.


Hi-cube containers are taller than standard containers at 9 ft 6 in (2.89 m). Hi-cube containers look distinctly taller than they are wide. By the end of 2013, half the word’s maritime fleet was in 40 ft hi-cube containers.

Other container sizes

In 2003 the EU commenced a process of defining the European Intermodal Loading Unit (EILU). The size of this container has yet to be defined, some 17 years after its proposal. It will be longer than 45 feet and, at 2.5 m about 10 cm wider than ISO containers. It will be difficult to fit it in with existing container ships. As can be imagined, the maritime freight organisations think it’s not a good idea.

From ship to rail and the problems of loading gauge

The size of freight allowable on trains is determined by what is called the loading gauge. European railways have loading gauges which are large and the transport of hi-cube ISO containers on such trains is no problem. Most of the British railway system is constructed to a loading gage called W6a. This is too small for the carriage of ISO containers.
W10 loading gauge allows standard wagons to carry ISO hi-cube containers. W12 allows room for these containers to have refrigeration packs as well. All new rail structures are built to W12 loading gauge.
There is a steady programme to enlarge UK railways to W10 and W12 loading gauge on the main rail freight routes so that containers can be offloaded from ships and then loaded directly onto rail lines for transport around the country but, as this map shows, only a small proportion of UK railways can carry the containers in which most goods are shipped.

How much does it cost to ship a container?

The website gave me a price of $5,500 (£4,100) for the cost of transporting a 40 ft hi-cube container from Hong Kong to Felixstowe UK on 5 January 2021. The shipping time quoted was 31-37 days. tells me that this journey, via the Suez Canal, is 11,047 nautical miles and at a speed of 13 knots will take 35.4 days.,hong-kong/port-of-felixstowe,united-kingdom/

How much fuel is used to bring our container to the UK?

As the following link shows, 13 knots is a particularly slow speed. A normal speed is over 20 knots. I’m going to use this figure to read off a fuel consumption for a contain ship carrying our container.

The graph says that a ship of 10,000 TEU will consume about 180 t per day. To calculate the worst case scenario, I am going to assume that the ship still takes 35 days to travel from Hong Kong to Felixstowe.

So total fuel used is 180 tonne/day x 35 days = 6300 tonnes.

The ship has a capacity of 10,000 TEU, and since our container is 2 TEU, there are about 5000 containers on the ship.

The amount of fuel used to transport our 2-TEU container is 6300/5000 = 1.2 tonnes. These tonnes we are using are metric tonnes, giving the confusing symbol mt in all the fuel tables. Current prices are $369 (£277) per tonne. That means that the fuel cost of getting our container from Hong Kong to the UK is 1.2 tonnes x £277/tonne = £332.

Bearing in mind that fuel prices were about three times as high three years ago, the fuel cost then could have been as high as £1000 and perhaps we would have paid extra because of that, perhaps an extra £900, making the total cost £5000.

In normal (non-covid) times, to get our container from Hong Kong to the UK would be about £5000, of which £1000 is the fuel cost.

What can we fit into a 40 ft hi-cube container?

An ISO container is 6.2 m x 2.43 m x 2.89 m externally. We need to make allowances for the thickness of the walls which I am going to suggest are each 0.125 m thick.

So the internal dimensions of the ISO container are 5.95 m x 2.18 x 2.64 m = 34 m³. This is 34,000 litres.

A shoe box for men’s shoes is 34.5 cm x 22.5 cm x 13 cm = 10,091 cm³ = 10 litres. We can fit 3500 shoe boxes into a hi-cube container but that is if they are sent in the box. If they are bagged and packed more efficiently, over 5000 pairs of shoes will fit into a hi-cube container.

Since it costs £5000 to bring 5000 pairs of shoes to the UK, that’s £1 per pair of shoes on transport cost of which 20p is fuel cost.

What’s in it for the ship owner?

In March 2010, the average price for a 10,000 TEU ship was about £90 million. That’s about £130 million in today’s prices.

As we’ve seen, the ship can transport 5000 containers, each size 2 TEU, from Hong Kong to the UK for a cost of £5000 each. But for each container it needs fuel worth about £1000, so the income generated by the trip is £4000 for each container x 5000 containers = £20 million.

The trip takes around 35 days, so the ship can do 10 such trips per year. If each is as profitable, that is net receipts of £200 million per year.

In rough figures, buy a ship for £130 millon. Earn £200 million a year carting things around the seas of the world with a profit margin of say 10%. So you earn £20 million a year and pay off the ship after 6 or 7 years. This is obviously a rough guess at the sizes of the figures involved.

Container ships are on average the newest ships at sea with an average age of 9 years. This is because so many have been built in the last couple of decades with the rapid expansion of world trade.

Bases for GXS53 lamps

Fixed holders for B22 are commonplace. Generally they are ugly fittings, used simply to provide light – cupboards (particularly under stairs), WCs (particularly in commercial premises and schools), etc. They are available as batten holders, on roses with provision for 3-plate wiring and in angled versions.

Why are they used? Because they are cheap, convenient, omnidirectional and have replaceable lamps.

What’s wrong with them? High profile, mechanically fragile, touchable contacts.

What should we have instead?
GX53 equivalents: batten, on 3-plate rose and on single face plate.

Why GX53?
Mechanically robust, concealed contacts, for ceilings where you don’t want to pierce the construction for thermal, fire and acoustic reasons and to enable easy conversion from existing B22 ceiling roses. Plus cheap, convenient, omnidirectional and replaceable lamps.

GX53 is large enough to have photocells and PIR sensors built while retaining the low profile.

There is scope for large and attractive lamps, for instance this from Fumagalli:

Is there nothing already available? Generally no. Some are available for recessed mounting. But those for surface mounting are crude, leave connecting wires visible and have no space for 3-plate connections.  

How much does it cost to run a 1 kW heater?

Here’s something terribly old-fashioned. It’s a three-bar electric fire. When all three bars on it is rated at 3 kW (3 kilowatts) but with only one bar, the middle one at the moment, its power is 1 kW.

I chose to check the power of the fire by using plug-in power meter. Here is the reading.

You can see it says 1061 W, that means 1061 Watts, just over 1000 watts, or just over 1 kW.

It is useful to know that a watt is a joule per second. If an appliance, like this electric fire, has a power of 1000 watts that’s exactly the same as saying that it has a power of 1000 joules per second. It means that every second the appliance takes 1000 joules of energy from the supply and, in this case, delivers it into the room, using it to raise the temperature of the room and make up for the heat loss.

Just for some fun, let’s calculate how many joules this electric fire takes from the supply in an hour.

The fire takes 1000 joules every second.

In 1 minute there are 60 seconds so this fire will 1000 x 60 = 60,000 joules.

In an hour there are 60 x 60 = 3600 seconds. So this fire will take 1000 x 3600 = 3,600,000 joules.

Here’s another way of working the numbers out. The power of the fire is 1 kW. When it’s on for an hour the energy it takes is 1 kilowatt x 3600 seconds = 3600 kilojoules. Yes, that 3600 kilojoules is the same as the 3,600,000 joules we have just calculated.

What we have done to calculate the energy supplied to our fire is multiply the power of the fire by the time for which it is on.
Energy = power x time. That’s a formula that is worth remembering.

There is another unit of energy, a convenient one, that we use when we want to measure energy. The unit is the kilowatt hour. Let’s use our new formula to calculate the energy drawn by our electric fire running for 1 hours.

Energy = power x time = 1 kilowatt x 1 hour = 1 kilowatt-hour, abbreviated to 1 kWh.

The kilowatt-hour, or kWh, is the standard unit for measuring electricity supplied. Indeed it is so standard that it is used for gas and sometimes other fuels as well. It is even often referred to as ‘the unit’.

At the time of writing, October 2020, energy from the electricity supply costs just under 15p per kilowatt hour and, to make numbers convenient, I’m going to use 15p as the price of a unit.

Our 1 kW electric fire uses 1 kWh of electricity every hour. And since electricity costs 15p per kWh, the fire costs 15p per hour to run.

What does it cost if it is on for a whole year.

In a day there are 24 hours. In a year there are 365 days. So there are 24 x 365 = 8760 hours in a year.

If our electric fire is on for a whole year,
Energy = power x time = 1 kilowatt x 8760 hours = 8750 kWh.

(If some of this is used to you, try saying the unit ‘kilowatt hours’ to get familiar with it.)

What’s the cost to run this fire for a year? Well each kWh of electricity costs 15p.
So 8760 kWh will cost 8760 kWh x 15 p/kWh = 131,400 p, which is £1314.

Quick trick for working out the cost for a year’s continuous use

Let’s just remind ourselves of where we have come to. A 1 kW fire, that’s a 1000 W fire, costs about £1300 to run all year. This can be used to provide us with a very useful figure to remember.

If a 1000 W electrical appliance costs £1300 to run for a whole year, a 1 W appliance costs £1.30 to run for a whole year. Remembering this fact is jolly convenient for calculating the cost of running all sorts of things that are on for a long time.

A modern middle-sized LED light bulb has a power of about 8 watts.
Remembering that a 1W appliance costs £1.30 to run all year, an 8 W led lamp costs 8 x £1.30 = £10.40 per year to run continuously. If it only runs for 3 hours, that’s an eighth of a day, then it costs £10.40/8 = £1.30 to run every day for 3 hours.

But an old-fashioned hot filament light bulb would need about 60 W for the same brightness. 60 W would cost 60 x £1.30 = £78 a year to run continuously, or £9.75 (£78/8) to run for 3 hours every day for a year (1/8 of a day).

How much does it cost to run our electric fire for an hour?

Just to remind ourselves, let’s calculate how much our 1 kW electric fire costs to run for 1 hour.
Energy = power x time = 1 kW x 1 h = 1 kWh. That’s pretty obvious but it’s worth remembering where it came from and how to check the calculation.
And, since energy from the electricity supply cost 15p/kWh, the electric fire costs 15p per hour to run.


This post is all about heating, so we are only going to consider heating appliances. It’s very easy to use electricity efficiently for heating. That electric fire is 100% efficient. That means that if it takes 100 joules in from the supply, it gives 100 joules out to the room. But gas heaters are less, sometimes a lot less, than 100% efficient.

The price of gas

Again at the time of writing, October 2020, gas prices are very low, less than 3p per kWh. That is in an economic climate in which much of the world is in lockdown and fossil fuel prices are really very cheap. A more realistic price long term price is nearer 5p per unit and I am going to use this as the figure for my calculations.

Gas fires

Gas fires are not nearly so efficient as electric fires as the following data will show. And, as we shall see, they are generally much less efficient when they are not on full output.

Here are the specifications for the above Flavel fire.

If your maths is quick, you might question have picked up something funny about these specifications. This fire gives 2.7 kW output for 6.5 kW input.
Efficiency = output/input = 2.7 kW/6.5 kW = 42%. This is much below the 56% that appears on the manufacturer’s specification. This is because manufacturers use what is called gross efficiency, a higher figure than the net efficiency we have calculated. This is explained later. For the moment we’ll run with the 42% efficiency.

If we turn this fire down a bit from 6.5 kW input, and if it has the same efficiency, (pretty unlikely because gas fires become much less efficient when used at low power), if the input is 2.4 kW, then the output is 2.4 kW x 42% = 1 kW. That’s the same figure as our electric fire. But to give out 1 kW heating the room the electric fire took in 1 kW. To give out 1 kW to heat the room, the gas fire needs 2.4 kW.

Where does the wasted energy of gas fires go?

If you’re a heating engineer, you’ll know the answer. For the average consumer it’s not so obvious what happens to the energy wasted by a gas fire. The answer is that it goes up the chimney and heats up the atmosphere. When your gas fire is on low, and taking 2.5 kW from the gas supply, 1 kW comes out into the room and 1.4 kW is wasted going up the chimney.

Hourly cost of a gas fire

On low, our gas fire takes in 2.4 kW from the gas supply (to give us 1 kW into our home).

Energy = power x time = 2.4 kW x 1 h = 2.4 kWh. Since gas costs 5 p/kWh, our gas fire costs 2.4 kWh x 5 p/kWh = 12p to run for an hour. Sure it’s a bit cheaper than an electric fire but not very much and, as we see later, gas fires are very inefficient at low powers and the real cost will be more than this.

More efficient gas fires

Here is a more efficient gas fire.

And here are its specifications:

Let us again calculate its efficiency from the input and output figures.
Efficiency = output/input = 4.0 kW/6.5 kW = 62%. Again you can see that this is more than the specified 66.8%.

You can see that this fire gives out by both radiation and convection. The fire has a box around the hot bits which air from the room can circulate through and take more of the heat from the burning fuel leaving less to be wasted up the chimney.

Let’s see how much it costs per hour to match the electric fire.
For 1 kW output it needs 1.6 kW input.
Energy input = power x time = 1.6 kW x 1 hour = 1.6 kWh.
With gas costing 5 p/kWh, the cost per hour = 1.6 kWh x 5 p/kWh = 8p.

Even more efficient gas fires

Here is Flavel’s most efficient gas fire.

Again let’s look at its specifications:

Calculating again the efficiency from the input and output figures,
Efficiency = output/input = 3.4 kW/4.5 kW = 76%, which is again below the 84% quoted by the manufacturer.

Again you can see that this fire heats the room by both radiation and convection. It has a box heat exchanger around it through which the room’s air flows, extracting more heat from the flame. But there’s another difference compared with the previous two fires. This fire has a glass front which restricts the amount of air going up the chimney. Excess air going up the chimney increases the losses to the atmosphere. The glass front massively removes this source of waste.

Let’s see how much it costs per hour to match the electric fire.
For 1 kW output it needs 1.3 kW input.
Energy input = power x time = 1.3 kW x 1 hour = 1.3 kWh.
With gas costing 5 p/kWh, the cost per hour = 1.3 kWh x 5 p/kWh = 6.5 p.

So an efficient gas fire would seem to cost less than half that of an electric fire for the same output.

Gas fires are very inefficient at low outputs

Here is detailed information about the Flavel Kenilworth High Efficiency fire discussed above.

You can see that at minimum heat output the efficiency of this fire is only 50% at low outputs. That’s the manufacturer beneficial efficiency figure. The net efficiency (the real one we have been using) is therefore about 45%. So to get 1 kW out the fire will take in 2.2 kW. This will cost 11 p per hour to run, not very much less than the 15 p for the electric fire.

If you have an inefficient gas fire, like the first on our list, at low outputs its efficiency may well be only 33%. So to get 1 kW out it will need 3 kW in. Therefore its running costs will be 15p per hour, just the same as an electric fire.

Is the 5p per kWh figure for gas fair?

Historically gas prices have been over 4p per kWh and my guess is that they will return to these levels. Your guess, whatever it is, may well be better than mine.

Why are the quoted efficiencies of gas fires always more than our calculations?

You will not be surprised to hear that manufacturers put the best spin on their figures. They therefore use an efficiency figure that makes their fires look as efficient as possible. Over the last 30 years they have moved from what is called net efficiency to gross efficiency.

When gas is burned, water vapour is produced. Gross efficiency assumes that the energy contained in the water vapour cannot be recovered, whereas net efficiency assumes that it can be recovered. In fact we know from condensing boilers that it is possible to recover energy from the water vapour produced when gas is burned.

Have a look at the (planned to be next on the list) post about condensing boilers and efficiency for full calculations of efficiency when the water vapour is allowed to condense.

Best plasterboard fixings

This is a condensed summary of a useful but very long YouTube video, the link to which is at the bottom.

Four types of plasterboard fixing were tested. Here are the resulting weights held when a single one of each type was used to secure the top of a shelf bracket.

Cast aluminium screw-in, 12p (pack of 100) Screwfix, less than 10 kg.

Fischer UX6, 6p (pack of 100) Screwfix – 12.5 kg.

Gripit, 52p (pack of 25) Screwfix – 20 kg.

Spring toggle, 45p (pack of 20) Screwfix – 22.5 kg.

Easyfix Hollow Wall Anchors, 13p (pack of 100) Screwfix – 22.5 kg.
Note that a Rawlplug version is 30p each.

Further comment

On another YouTube video a variety of plasterboard fixings were demonstrated by a different method to the above. The results are below, but there is a pattern evident. There are fittings which are strong: wall anchors, spring toggles, etc. These are twice as strong as fixings that are weak: metal plasterboard screws, pretty much all the ones that get a good grip of a reasonable area of the back of the plasterboard are pretty similarly strong, but only 2 or perhaps 3 times the strength of those things that are obviously much weaker.

GeeFix £2.23 each Total failure weight in 12.5mm plasterboard was 125KG or 275Lbs Hollow wall anchors £0.50 each Total failure weight 109KG or 240Lbs Snap toggles £1.30 each Total failure weight 129KG or 284Lbs Blue GripIt £0.83 each Total failure weight 101KG or 222Lbs Snap toggles £1.30 each Total failure weight 129KG or 284Lbs Spring Toggles £0.54 each Total failure weight 176KG or 387Lbs Metal plasterboard screw £0.37 each Total failure weight 53KG or 116Lbs


Use the UX6 and put twice as many in. That gets you greater strength than any of the others for a price almost identical to the cheapest.

My own experience is that standard Rawlplugs, either the 6 mm or 7 mm provide quite a strong fixing in plasterboard provided that you drill a very clean hole with a normal twist drill (not a masonry drill because these give a more ragged hole). At 2p a time for the brown ones they are hard to beat but I am not aware of any strength tests.

How Geohash works.

Geohash is a way of referring to any point on the earth with a single string of characters. It was invented in 2008 by Gustavo Niemeyer and GM Morton.

Geohash first splits the equator into 8 sections, each 45° of longitude wide. Then it splits the latitude into 4 sections, again each section 45° of latitude wide. This splits the world’s surface into these 32 regions.

These 32 regions are labelled, using the 10 digits 0-9 and 22 letters (omitting A, I, L and O).

It is easier to see what is going on if we draw these regions on the sort of flat map we are used to.

The single character Geohash reference 6 identifies much of Latin America. Australia spans codes Q and R and China is mostly code W.

For more precision we subdivide each of these large regions into smaller regions. Here we go down to level 2.

At level 3 you can see that Mysuru is in square tdn.


With 9  characters, location can be specified to just under 5 m, which is sufficient for most purposes. But with up to 13 characters, you can specify a location to within less than 5 mm.

How subdivisions are carried out

Level 1 splits a rectangular map of the earth into 32 rectangles 8 rectangles wide and 4 high. The splitting from Level 1 to Level 2 subdivides with new rectangles though this split is 4 rectangles wide and 8 high.

Subdividing from Level 2 to Level 3, the rectangles revert to 8 wide and 4 high. This alternating splitting is maintained throughout the levels. The result is that passing through 2 subdivisions a given area is split into 32 x 32 (ie 1024) subdivisions.

Z-shaped region labelling

Geohash regions are labelled in a Z-shaped order as shown.

This give the best fit in terms of closeness between the Geohash, (a one-dimensional structure) and the surface of the world (which is two-dimensional).


The QALocate system is a brilliantly simple way of referring to locations, buildings, routes and structures. At its heart is a way of identifying any point on the earth’s surface using a human-friendly sentence-like group of words.

For instance New York is xxxxxxxxxxxxxxxxxxx; London is xxxxxxxxxxxxxx, Tokyo is xxxxxxxxxxxxxxxxx and Delhi is xxxxxxxxxxxxxxxxxxx.

Qalocate can be as precise as you like. The front door of the white house is identified to within 150 mm by the phrase: xxxxxxxxxxxxx. If necessary locations can be expressed to within 5 mm by a seven-word phrase.

(This needs a link to a map where you can get to your location, click on it and produce a QAcode.)

The core of QALocate is simple to use and understand. It is open source, meaning that anyone can use the system completely free of charge. QALocate is mathematically simple, so translation to and from QALocate references is easy: coding and decoding software can incorporated into the simplest of electronic devices.

As well as identifying locations, QALocate comes with other powerful tools.

PointCodes, a “!” followed by up to 63 characters, the simplest way for anyone to refer to a location or structure.

StrutureLocator, a globally unique, alphanumeric identifier for every building and structure on the planet.

Location Naming System (LNS), turning human-friendly names to StructureLocators.

Waysette, a navigation app purpose-built for rideshare drivers.


How do four-word QALocate phrases work?

QALocate builds on the equally brilliant Geohash method of identifying points on the earth’s surface. (See our explanation of how Geohash works.)

The 9-character geohash for New York is dr5regw3p which identifies a square 4.7m x 4.7 m right in the centre of New York. (Should be specified.) For the populated regions of the world we encode this 9-digit geohash into human-readable form by splitting it into four chunks and encoding each as a word.

dr5        re         gw        3p

Man     bites    angry    dog.

Noun   verb     adjective  noun

The choice of noun-verb-adjective-noun produces sentence-like groups of four words that are easy for humans to communicate. The choice of words is such that minor inaccuracies common in communication do not affect the outcome: ‘men bit angry dogs’ translates to the same geohash because QALocate treats singular and plural as the same and every form of the verb as the same.

Encoding and decoding are both a simple process of looking up in a small group of tables to find the word that corresponds to 2 or 3 characters, or the 2 or 3 characters which correspond to a word.

 Why do we need five-word phrases?

Four-word phrases cover the populated areas of the world but there are insufficient familiar words to cover the whole world with four words. QA locate uses five-word phrases to extend its coverage over the whole world, adding one of 32 adjectives to identify the first letter of the geohash.

Hairy goat eats thin camel.

Again the phrases are sentence-like and easy for humans to remember.

How accurate can QALOCATE specify

QALocate can specify to a precision of less than 5 mm by using phrases up to 7 words long.

Examples here.

What’s wrong with Open Location Code

Lift much of existing text.

Why QALocate beats What3Words.

Text required: no proprietary, only 7 lookup tables, similar codes are near and, generally, near things have similar codes.

Where does the Geohash come from?

Geohash first splits the equator into 8 sections, each 45° of longitude wide. Then it splits the latitude into 4 sections, again each section 45° of latitude wide. This splits the world’s surface into these 32 regions 

It is easier if we draw these regions on, as we are used to, a flat map which represents the earth’s surface.

The single character Geohash reference 6 identifies much of Latin America. Australia spans codes QR and China is mostly code W.

For more precision we subdivide each of these large regions into smaller regions. Here we go down to level 2.

Here is level 3 and you can see that Mysuru is in square tdn.

With 9  characters, location can be specified to just under 5 m, which is sufficient for most purposes. But with up to 13 characters, you can specify a location to within less than 5 mm.

Geohash is a 32-base system. It uses the 10 digits 0-9 plus 22 alphabet letters, omitting letter A and the letters I, L and O which are easily confused.

Luminous efficacy — how cheaply you might light your house one day.

Where do we get light from?

In the beginning there was only one source of light – the light from the sun. That light is electromagnetic radiation, emitted by the sun, a very hot body. At first all artificial light sources mimicked the sun because they, too, were produced by hot bodies – by fires, by burning oil or burning candle wax. Then along came electricity which also mimicked the sun by passing an electric current through a wire to make the wire hot enough to give light out.

Light output of lamps

A most important quality of a lamp is how much light it gives out. We are not, yet, very much used to measuring the light output of lamps. In really quite recent days we all bought the same type of lamp, the hot wire lamp, called an incandescent lamp. We used to measure the brightness of a lamp by the power we supplied to it. An average sort of brightness was a 60 W bulb. A much brighter bulb was 100 W and a dim one perhaps 40 W or even 25 W.

Nowadays, with different types of lamps needing different powers, we need a unit that tells the actual light given out. That unit is the lumen. The old 60 W lamps gave out about 800 lumens. A brighter 100 W lamp gives out about 1500 lumens and a 40 W lamp about 500 lumens.

You know that hot bodies used for lighting are white hot. The white light they give out isn’t a single colour. It’s made of a spectrum of colour, with light from deep red to violet. To see properly we need light of all the different colours. That’s why we use white light for lamps.

Luminous efficacy

The luminous efficacy of a lamp tells you how much light you get out (in lumens) per watt of power you put in. Lamps with a higher efficacy give you more lumens per watt. It is as simple as that. The old-fashioned tungsten filament incandescent lamp gave out about 800 lumens (800 lm) for an input of about 60 watts (60 W);
that is (800 lm)/(60 W) = 13 lm/W or 13 lumens per watt.

The most efficient light would be yellow-green

The eye is the most sensitive to this yellow-green light which has a wavelength of 555 nm. It’s this sort of colour.

Image for post

You can refer to this site if you want to see the colour of different wavelengths of light.

A lamp that gave out light of only this yellow-green colour would be the best for brightness. If you had a device that converted all its input power into light at that yellow-green wavelength, you’d get a luminous efficacy of 683 lumens per watt. That means that, if you were happy with yellow-green, you could have a light bulb, roughly the equivalent of an old 60W bulb, needing only 1 W to power it. Of course that would be bright but it wouldn’t be any use for many purposes because everything would look strange different shades of yellow-green. What we need is a white light that has a range of colours in it, even if our eyes are not as sensitive to other colours.

Hot bodies produce a whole range of radiated energy, only some of which is visible.

Let’s have a look at the range of wavelengths of electromagnetic radiation given out by a hot body. A hot body near the melting point of tungsten (3695K) produces some light at 555 nm and the rest over other parts of the electromagnetic spectrum including lots of UV and IR.

Image for post

A black body at the temperature of melting tungsten, somewhere between 3500 K and 4000 K, giving all its input out by electromagnetic radiation (that means no energy conducted out through the connecting wires, etc), could have an efficacy of 55 lm/W. That’s a lot more than the efficacy of the old fashioned 60 W hot-wire light bulbs which we calculated had an efficacy of 13 lm/W. That’s because it has to operate at a temperature much lower than the temperature at which its filament would melt.

If you could find a material that would survive being run 7000 K, twice the melting point of tungsten, hotter than the surface of the sun, that light could give out 95 lumens for each watt input, an efficacy of 95 lm/W. For many years, and still now in some cinemas and sometimes in lighthouses, arc lights are used. These make use of extended sparks through a gas from one terminal a lamp to another. So the ‘filament isn’t a solid that could melt, it is a gas.

LED lamps avoid giving out light you can’t see.

Modern lamps are made with Light Emitting Diodes (LEDs). LEDs use a different technique to make a high efficiency light bulb. With LEDs you can build a device that only gives out the light you can see, so it doesn’t waste electricity giving out ultraviolet or infra red light. If you chop out the hardly visible colours but otherwise mimic a black body spectrum (the temperature of which doesn’t change the figures much) you can get 250 lumens of light out for each watt of electricity input.

If you reduce the spectrum of your light, so your LED does not give out light where the sensitivity is down to 5% of the maximum, you can get and even higher luminous efficacy – 350 lumens for each watt, though obviously your light doesn’t show colours so well. We say that there is some deterioration of colour rendering. If you keep light down to the 2% threshold, you can get an efficacy about 300 lm/W with reasonable colour rendering.

Back to the beginning and the extreme, if you only give 555 nm light out, you could get to 683 lm/W.

The best available lamps nowadays.

The best lights I have seen so far for domestic lighting are ones like these sold by Argos for £2.50 each.

Buy Argos Home 6W LED ES Light Bulb – 2 Pack | Light bulbs | ArgosAn impressive A++ energy rating, an Edison screw fitting, 6 watts and 806 lumens, this pair of bulbs are ideal for…

This Argos lamp gives 806 lumens for 6 watts of input, which is 134 lm/W.

Do you remember from the beginning where I said that a 60 W lamp gives out about 800 lumens? The Argos LED lamp gives out the same light as an old-fashioned lamp but only use 6 W instead of 60 W. Its efficacy is ten times more than that of a hot wire lamp.

Here is a brighter lamp.

Image for post

This lamp gives out 1521 lumens for an input power of 12 W.
That is 1521/12 = 127 lumens per watt. This lamp is available from and is also dimmable.

The energy balance for a lamp

We’ve just said that the maximum theoretical efficacy you can get for any lamp is about 300 lumens per watt. At the theoretical limit, all the energy coming out of the lamp would be electromagnetic radiation tailored to match a hot-wire lamp but with the spectrum trimmed to cut of light outside the visible spectrum that had no lighting effect.

Just imagine a perfect bulb gving out a curtailed spectrum of 806 lumens with an efficacy of around 300 lm/W.

That lamp would be giving out
(806 lm)/(300 lm/W) = 2.7 W of light.

So a perfect 800 lumen lamp would need 2.7 watts. The Argos lamps above takes in 6 W to give out 2.7 W of light, so they are wasting 3.3 W, which is being given out as heat.

Let’s compare that with an old 60 W tungsten lamp which also gave out about 800 lumens. The hot-wire tungsten filament lamp used to give out the same 2.7 watts of light as the LED lamp. But it was taking in 60 W.
So it is giving out 60 W – 2.7 W = 57.3 W of heat.
It’s not surprising that the old tungsten lamps were hot – taking in 60 W of light, giving out 57.3 W of heat and only 2.7 W of light.

How far is there to go?

Heating effects of currents in wires and semiconductors are bound to limit efficiencies, let’s say to 70% of theoretical. That means that we could guess that the maximum efficacy a lamp could get might be about 200 lumens per watt. It certainly couldn’t be more than 300 lumens per watt.

IntegralLED, who make the 1521 lumen lamp mentioned above, have a ceiling panel lamp that has an efficacy of 152 lumens per watt.

Image for post×600-23w-4000k-3500lm

The vendor of LED lamps at this website says that some of its products have efficacies of up to 170 lm/W.

So LEDs are well on towards the efficiency that any lamp could possibly have. What this means is that the 60 W, or should I say 800 lumen, lamp which you can buy at Argos for £2.50 and consumes about 6 W could, with today’s technology, be made to need only 5W. In the future we could see that dropping to 4W or so, but the theoretical limit is to something over 3W.

More efficient lighting is the main reason our electricity production has reduced.

The number of homes and the number of people in the UK have steadily increased over the years. Yet the amount of electricity used for lighting has gone down by a third since 1997. Here you can see how the amount of electricity generated has decreased over the past thirty years.

Image for post

The UK population rose from about 58 million in 1997 to about 68 million today. With more people, more homes and more electrical appliances, electricity supplied grew, as you can see from the graph above. But less electricity was needed after 2005. This is because about a quarter of our electricity supply was used for lighting, and now that lighting is much more efficient, we don’t need so much electricity.

What will lamp manufacturers sell when we all have LED lamps?

The Argos lamp referred to above has a quoted life of 15,000 hours and 100,000 switching cycles. That light should last 10 years at 4 hours a day. This Ryet bulb from Ikea has a lower luminous efficacy. It gives out 470 lm for an input of 4.5 watts; that’s 104 lm/W. But it only costs £1.

Image for post

It’s the ultimate consumer product: costs nothing to buy, nothing to run, needs no maintenance and lasts for ever. But what will lamp manufacturers sell when we all have LED lights. For a while they can persuade us to trade up to ever more efficient lamps but there is not much saving left to be made. After all, the 60W lamp we used to buy only uses 6W now and could never use less than 3 W.

Perhaps then the manufacturers will make LED lamps that fail early. After all, that’s what they do with cars. Cars could be designed to last for twenty years or more but instead they are designed to fail in 8 to 10 years so we go out and buy new ones.

In 1971 I had a tour of the physics department of Durham university. They showed me a new component. ‘It’s a diode really but it gives out light when it conducts. It’s called a light emitting diode or LED for short. We think it might be useful.’