Ira Edward Aaron, Ph.D., 1919-2016

My great-uncle, Ira Edward Aaron, Professor Emeritus of Reading Education at the University of Georgia, died on September 27 at age 97 after a long illness. He was well taken care of by my sister and died in bed at her house. Thankfully, all the way to the end, his mind was clear when he was awake, though he was awake less and less of the time.

My middle name, Aaron, was chosen to honor him and his branch of the family.

"Uncle Ed" was an important role model for me. He was the first academic in our extended family. My first visit to the University of Georgia was when, at age 6, I was brought to his department to be studied. Normally they evaluated children to find out why they couldn't read — they evaluated me to find out why I could, and how I had learned it before starting school, and to make recommendations for my further education.

He was a co-author of many editions of the Scott, Foresman elementary-school reading books, including Fun with Dick and Jane, so a whole generation of Americans grew up reading things he wrote. He boasted of being co-author of one of the most notorious sentences in American literature, "See Dick run."

A sample of "Uncle Ed's" academic work can be seen in this pamphlet for teachers published by the State of Georgia. Notice how clearly he writes. Almost all his academic work was applied rather than theoretical; his emphasis was on getting the word out, helping teachers everywhere teach more effectively. During the short times our careers overlapped, we never collaborated on anything, which may have been a lost opportunity. I did give a talk to his department while I was a Ph.D. candidate; I talked about the writing systems of other languages and what was involved in learning to read them.

He had the distinction of being listed on page 1 of Who's Who in America and occasionally getting mail for the baseball player Hank Aaron, whose entry was right above his.

He is not to be confused with at least two distant relatives also named Ira Edward Aaron, one of whom was Ira Edward Aaron, M.D., of Lyons, Georgia.

Update: Dr. Aaron's colleague Donna Alvermann has collected tributes from former students and colleagues and I have also added a scanned image of the memorial card that was given out at the visitation. More material may be added here later.

Permanent link to this entry

Will the real M103 please stand up?

Here's a pair of star clusters in Cassiopeia:

Stack of four 1-minute exposures with the AT65EDQ refractor and Nikon D5300. Now here's the same image, "solved" (identified) by Astrometry.net:

One of those star clusters is Charles Messier's M103. But Messier did not give an exact position for it. Back in 1832, without benefit of modern star maps, W. H. Smyth identified M103 as the cluster later designated NGC 581. But in Sky and Telescope, February, 2016, p. 8, Tom Reiland argued that NGC 663 is a better candidate. Both are consistent with Messier's description, which was a report of an object discovered by his colleague Pierre Méchain with no exact position given.

But I don't advocate changing the designation now. Too much confusion would result.

By the way, don't confuse Admiral William Henry Smyth, navigator, astronomer, and historian, with Lord of the Admiralty William Henry Smith, unpopular politician. They were contemporary.

Permanent link to this entry

Closer to earth

Taking a break from astronomy today, here are a couple of nature photos. The crape myrtle is encroaching on our driveway and will soon be cut back, but it has been a pretty sight all summer. The lizard (Anolis carolinensis) was actually climbing a wall; the picture has been rotated 90 degrees.

Nikon D5300, handheld, with 18-55mm kit lens.

Permanent link to this entry

Barnard's Galaxy, NGC 6822

You can't see the Magellanic Clouds from our latitude, but here is the next best thing.

The Magellanic Clouds are small irregular galaxies that are satellites of our own, and from Australia, they look like detached pieces of the Milky Way. Barnard's Galaxy is a similar irregular galaxy, not quite so close. It's at roughly the same distance as the Andromeda Galaxy (about 2 million light years), and because it is nearby and not very densely packed, it is one of the first exterior galaxies in which Edwin Hubble studied variable stars. Until he found evidence for its distance, he thought it was a large star cluster, which, in some sense, every galaxy is.

Now look 2/3 of the way from the galaxy to the upper right corner of the picture. You'll see two bright stars. One of them looks a bit greenish; it is actually the planetary nebula NGC 6818.

The picture is grainy because of the contrast amplification that I had to perform in order to bring out this faint object in a picture taken in town, through rather hazy air. Stack of 20 1-minute exposures with the AT65EDQ and a Nikon D5300 at ISO 400. I hope to rephotograph Barnard's Galaxy from a dark-sky site.

Permanent link to this entry

In the heart of M24

I know this picture won't fit on your screen all at once, but scroll through it and enjoy the view...

Digitally processing film astrophotos

Kingdom of the black 2012 Ford Escapes

When I bought a black 2012 Ford Escape five years ago, I had no idea everybody in town was doing the same thing.

Today on the way to the University I encountered three of them. It is very common for me to find myself parked near, or even next to, another black 2012 Escape.

The 2012 Escape is easy to recognize because it has the body style of the 2008-2011 Escape with distinctive taillights, which were used only that year. Black is the most common color. Most don't have quite as much silver on the lower front grill as the one in the picture, which shows an option.

I know that 2012 was the end of a body style that was being discontinued; the Escape was completely re-engineered for 2013. As I understand it, the Kansas City, Mo., production line was shut down at the end of the 2012 model year.

I also know that I got a good price on my Escape. Were black 2012 Escapes sold at clearance prices everywhere, or just in Athens, Georgia?

Permanent link to this entry

Lunar craters

These pictures were taken with my 8-inch telescope and Canon 60Da camera on September 14. Each is a stack of the best 75% of several thousand video frames, stacked with AutoStakkert and sharpened with RegiStax.

Pythagoras:

Grimaldi:

Seleucus (small and prominent, with Eddington beyond it):

Bailly:

Note how well the Canon renders the tonality and subtle coloration of the lunar surface, especially in the first picture.

Permanent link to this entry

Do you need a modified DSLR to photograph nebulae?

It helps, as you can see from the pictures, but it's not vital.

The picture on the left was taken with an unmodified Nikon D5300; the one on the right, with a Canon 60Da, which has a special filter that transmits more deep red light (at 656 nm) than an ordinary DSLR.

Thin hydrogen nebulae such as M8, which you see here, emit about 3/4 of their light at the wavelength of hydrogen-alpha, 656 nm. The other 1/4 is mostly at hydrogen-beta, 486 nm.

All DSLRs detect hydrogen-beta well, and they detect hydrogen-alpha to some extent, but not fully. Thus the nebula is purplish, a mix of red and blue, in the Nikon picture. The Canon camera responds fully to hydrogen-alpha, so the red color predominates and a bit more of the nebula is visible.

Back in the days of film, Ektachrome responded strongly to hydrogen-alpha, and other color films didn't, nor did most black-and-white films.

Today, Canon and Nikon both make special DSLRs for photographing nebulae; Canon's is the 60Da (used here) and Nikon's is the D810A. But the usual way to get a modified camera is to buy a conventional one and send it to LifePixel, Hap Griffin, or another independent modifier. What they do is replace the camera's infrared-cutoff filter with one that transmits more of the deep red end of the spectrum. After the modification, the camera is still good for other kinds of astrophotography and even daytime photography (with vivid Ektachrome-like colors).

Do you have to do this? No. As you can see, the unmodified camera performs only a little worse than the modified one. The difference is that the nebula isn't red. People who grew up in the Ektachrome era are sometimes very strident in their insistence that nebulae ought to be red. Not me... I just want to see the nebula. Seen in the telescope, it is bluish-gray!

Permanent link to this entry

NGC 663 a new way

This is a stack of twenty 5-second exposures of star cluster NGC 663 through my 8-inch telescope at f/10, using a Canon 60Da at ISO 3200. I'm experimenting with short exposures. If I can work with 10-second or shorter exposures, I won't need an autoguider, because that's how long the autoguider goes between corrections. Emil Kraaikamp, over in Belgium, has been photographing galaxies with hundreds of short exposures.

NGC 663 may be what Charles Messier saw when he recorded the object known as M103. Traditionally, M103 has been identified as NGC 581, but NGC 663 is more prominent. it's not clear from Messier's description which cluster he actually saw. See Tom Reiland's letter in Sky and Telescope, February, 2016, page 8.

My picture isn't great, but it at least shows that the technique works.

Permanent link to this entry

Checks you can't write on

Annoying discovery of the day: Some modern "safety" checks don't accept ink from some pens.

I recently wrote a few checks using a newly-opened batch of checks ordered from Checks In The Mail several months ago.

They repelled the ink!

I tried two fountain pens and a typewriter and got patchy results with all. Ballpoint ink, however, worked.

To be precise, the yellow safety stripes repelled the ink. Quick experiments with checks from other batches showed me that Checks In The Mail's blue stripes are less afflicted than their yellow stripes, and Harland's yellow stripes (from an old batch) less afflicted yet, though not perfect.

Fortunately, Checks In The Mail graciously gave me a full refund.

Harland, the previous supplier, no longer seems to be in the retail check-printing business — upon logging into their web site, I was redirected to another company with which I was not familiar. So I ended up going with Vistaprint, which has much lower prices, and carefully picking a design that didn't have stripes.

If the new checks turn out to be ink-repellent all over, I'm not sure what I'll do, but at least they don't charge very much (about $18 for 150 wallet checks, delivered).

Before anyone tells me fountain pen ink is an obsolete technology, let me add that so are paper checks!

More about this: It's not just a fountain pen problem; other types of pens are converting to water-based ink for environmental reasons, so I expect other people to encounter it more and more. The purpose of the yellow stripes is to keep people from altering checks; they discolor if erased, I'm told. My checks, what few there are, are to government agencies, charities, and businesses; there's little risk of anyone altering them.

Permanent link to this entry

Has Chomsky been blown out of the water?

Word on the street is that the linguistics of Noam Chomsky has been discredited by new discoveries.

Has it? Click here for my perspective.

Permanent link to this entry

Don't let the media control your mind

This Facebook posting was very well received, so I'm preserving it here...

FRIENDS: Please don't let the media (including Facebook) control all your thoughts and feelings.

I have more important things to do today than trade guesses about why Hillary Clinton stumbled or who tore down the flags at Occidental College.

So do you.

Plenty of journalists and gossips want to keep you angry with half-baked stories and one-sided accounts of the facts. It's because they want your attention.

Give your attention to things that deserve it. Make the world a better place by doing your job, taking care of your family, helping those around you, and solving problems that only you can solve. Not everything important is on Fox or Facebook.

Permanent link to this entry

Lagoon and Trifid

Now the grand finale from my trip to Deerlick with the 65-mm f/6.5 refractor and Nikon D5300 body. Here you see the star cluster M21 (at top left), the Trifid Nebula (M20), and the Lagoon Nebula (M8, at bottom). Stack of five 5-minute exposures at ISO 400.

Why are they purple? Because this Nikon has not been modified for extended red sensitivity, unlike my Canon 60Da, so the red hydrogen-alpha wavelength does not dominate the picture (though it is picked up to a considerable extent). Also because digital cameras pick up the blue emission of hydrogen-beta that film did not. On Ektachrome film, nebulae are red. To the eye, they are gray, pinkish, or even greenish. To digital cameras, they are often purplish.

Permanent link to this entry

On to Andromeda

When I test equipment, I see familiar objects with new eyes. This is with the same telescope and camera as the other pictures I've been showing you (below); 65-mm f/6.5 refractor, Nikon D5300, three 5-minute exposures at ISO 400.

This is of course the familiar Andromeda Galaxy, taken at the Deerlick Astronomy Village.

Permanent link to this entry

How to take a picture of the Milky Way

[Updated with better color balance.]

In the past couple of years, lots of nature photographers have been taking and publishing pictures like this one. I'm glad, not just because I like the Milky Way, but more importantly because I want people to appreciate the starry sky and the fact that excessive city lights keep you from seeing it.

How do you take a picture like this?

You probably need a DSLR or a pro-grade mirrorless camera; miniaturized cameras with tiny lenses and tiny sensors don't do as well. But try what you have.

First, find the Milky Way. The bright part of it is not always in the sky. At north temperate latitudes, look to the south, before midnight, from August to October.

To get a good view, you'll need to go out in the country. I took this picture at the Deerlick Astronomy Village.

Then put the camera on a fixed tripod. Focus on infinity, preferably by focusing on a bright star or very distant light; autofocusers can't see the stars. Some cameras have a mountain symbol that you can choose to lock the focus at infinity.

Expose anywhere from 3 to 30 seconds with a wide-angle lens wide open (f/4 or wider) and a high ISO setting. The longer the exposure, the more the stars will lengthen into short lines rather than dots due to the rotation of the earth.

This one was 6 seconds at ISO 12,800 with a Nikon D5300 and a 1990s-vintage Sigma 24-mm lens at f/2.8, with long-exposure noise reduction. Brightness and contrast were adjusted in Photoshop, where further noise reduction was done.

Permanent link to this entry

The h in Sarah (etc.) and the th in Esther

A friend asked me today whether the Biblical name Sarah was more correctly spelled with or without the final h.

The answer is that Sarah is like a lot of Hebrew names — it ends with an h that was never pronounced but was always written.

The original way of writing Hebrew did not include vowels, but it did use some consonants as hints of where vowels should be: y for e and i, w for o and u, and h at the end of the word only, for a. Later, when vowel symbols were added, these consonants were left in place.

That's how we got Sarah, Deborah, Isaiah, and all those other names that end in ah.

This led me to another question: Why is Esther written with th when it has never been pronounced that way? In Hebrew it is written as tav with dagesh, which is a T sound not softened as it would be between vowels.

The answer, as I finally found out a few years ago, is that because there are two T letters in Hebrew, tav and tet, it became traditional always to render tav into Greek as theta, and tet as tau, even though tav did not sound like a postclassical theta. (Note tav = theta and tet = tau. They were obviously going by the sounds of the letters at some early date, not the names of the letters, which are swapped.) We got our English spelling from Latin, which got it from Greek; theta always goes into Latin as th.

Permanent link to this entry

Scutum and a minor star cluster

Using the same setup as in the previous entry (below), here's a picture taken on the previous day, a stack of two 5-minute exposures of the heart of the constellation Scutum. It shows star clouds, dark nebulae (at the upper right), and the obscure star cluster Trumpler 34 at the bottom right. The crescent moon was in the sky.

Permanent link to this entry

Apo refractor moon

Can you photograph the moon with an apo refractor? Certainly. This is not a great photograph because the air was rather unsteady. On September 5, I took the refractor to Deerlick and got some excellent pictures, which you'll be seeing in the next few days. While out there, I snapped this 1/320-second exposure using a Canon 60Da camera body and "silent shooting" (vibrationless electronic first shutter curtain, a feature the Nikon D5300 does not have). Brightness adjustment and unsharp masking with Photoshop.

Permanent link to this entry

The holey star cluster M26

I'm experimenting with the setup shown below — a borrowed AT65EDQ 65-mm f/6.5 apo refractor, Nikon D5300 body, Celestron AVX mount, and a homemade autoguiding device similar to the now-discontinued SBIG eFinder riding piggyback on top, connected to an SBIG STV. Let me start with one of the less spectacular pictures; better ones are coming in the next few days.

The star cluster that you see above is M26, in Scutum, notable for appearing to have an empty space in the center. As I understand it, astronomers are still debating whether the "hole" is a dark nebula (common in the area) or an actual lack of stars.

This is a stack of five 5-minute exposures at ISO 400, taken in town (Athens, Georgia).

Finally, Peanuts readers will remember Charlie Brown's tree that ate kites. While completing the M26 exposure sequence, I caught a tree eating a star cluster:

Permanent link to this entry

Java under Windows Subsystem for Linux

I think I've found the Achilles' heel of Windows Subsystem for Linux in its present form: Java.

I haven't been able to get a Java system completely working under it, with either OpenJDK 7 (which installs if you ask for "default-jdk") or Oracle JDK 8.

The usual symptom is that a Java program hangs, completely occupying one CPU core. Depending on the circumstances, java -version may hang, or javac hello.java may hang.

But according to what I hear on the Web, even if you can compile and run simple single-threaded Java applications, you're not out of the woods. More complicated Java-based packages such as Bazel still have problems.

Why do I want Java under WSL when I can already run Java under Windows? Because some other Linux-based tools, such as the Bazel utility from Google, require it.

As best I and others can determine, the problem has to do with process creation. The biggest difference between Linux and Windows is that creating a process in Windows is a much bigger deal, taking more work and time. For WSL, Microsoft introduced a new type of lightweight process. It apparently still doesn't quite satisfy Java's expectations.

Permanent link to this entry

The midnight optician

Very late the other night, Melody's glasses broke. To be precise, one of the temples (earpieces) broke in two near the hinge.

Most people would have resorted to Scotch tape and an emergency trip to the optician the next morning, but we were more resourceful. Rummaging among Melody's old glasses, we found one whose temples were almost the same. Then I got out my jeweler's screwdrivers and made the swap. Success! A perfect fit; it didn't even need adjusting. But the repair job was non-trivial because the spring kept retracting the hinge where I couldn't get to the hole.

A very close observer may notice a slight difference in style between the left temple and the right temple of Melody's glasses. But, really, if you're going to stand that close to my wife, you might need to ask my permission!

Permanent link to this entry

Code signing: how I do it

A while back I wrote about code-signing certificates and how I was getting ready to start digitally signing my software products. Other things came up, and I didn't implement everything I wanted to until today. Now, EXIFLOG reaches you code-signed, and so does CoVec, or it will when I start selling it.

Recall that I got an Authenticode certificate from Digicert and installed it on my main desktop computer. I also got a Windows Software Development Kit (which comes with Visual Studio) in order to have the signtool command.

But there were tactical questions.

Sha256 or Sha1? Newer versions of Windows expect Sha256 encryption, which is more secure, but prior to Windows XP SP3 (or so I'm told), only Sha1 is acceptable.

One certificate or two? Executable exe files accept multiple digital certificates, but an msi file (Microsoft Installer) can only have one.

Access to signtool? It's available in the Visual Studio developer command prompt, but in an ordinary command prompt, it's not on the path and doesn't work.

My solution? First I used the command where signtool.exe (from a Visual Studio Developer command prompt, not an ordinary one) to find out where signtool resides. In my case, it was in
C:\Program Files (x86)\Windows Kits\10\bin\x64\
and I added it to the path.

Then wrote a script that includes the following three commands:

to affix an Sha256 signature, affix an Sha1 signature too (failing harmlessly if the file won't accept one), and then verify them.

Now then: Workflow with Visual Studio.

I'm well aware that I could call signtool as a post-build event and get my code signed automatically. I don't want to do that! I only sign code when, in my considered opinion, it is ready to distribute.

So here's what I do.

(1) Build the project (the executable).

(2) Sign the exe file. With Visual Studio installer projects, you have to sign the copy that is in obj\Release, not bin\Release. This threw me for a while. Installshield, however, uses the latter.

(3) Build (not rebuild) the setup project.

(4) Sign the msi file.

Permanent link to this entry

What if the Ross constraints are pragmatic?

OK, hang on to your hats, this Notebook entry is going to be about linguistics, on a rather technical level. But I'll do my best to make it understandable by non-specialists.

An article in the September 2015 issue of Language caught my eye. (Yes, I'm a whole year behind reading them.) Yusuke Kubota and Jungmee Lee argue that the Coordinate Structure Constraint is not what we thought it was.

The Coordinate Structure Constraint is an unusual, but strict, rule or principle of grammar that wasn't discovered until 1967 (by J. R. Ross). It will take me a moment to explain what it is.

First look at how the English language forms yes-no questions. We put do, does, or did at the beginning of the sentence and change the verb to the plain form (in this case see rather than saw):

You saw a duck.

Did you see a duck?

Now consider what questions. To form a what question, we make the changes that I just described, and we also put what at the beginning of the sentence and omit the noun phrase that is being questioned, thus:

You saw a duck.

What did you see _ ?

Here I'm using _ to mark the position of the omitted noun phrase.

Much of the early appeal of Chomsky's transformational grammar was that it enabled us to state recipes like this for making one kind of sentence from another. If you could account for the grammar of one set of sentences, you could make another set of sentences from them easily, by transformation, without having to describe everything from scratch.

Now note a crucial fact about what questions: The questioned (omitted) noun phrase can be almost anywhere in the sentence. Exactly one noun phrase can be questioned, but it can be almost anywhere.

You saw a duck carrying a stick.
What did you see _ carrying a stick?   (A duck.)
What did you see a duck carrying _ ?   (A stick.)

You saw a duck sitting next to a goose.
What did you see _ sitting next to a goose?   (A duck.)
What did you see a duck sitting next to _ ?   (A goose.)

You saw a duck carrying a stick shaped like a baseball bat.
What did you see a duck carrying a stick shaped like _ ?

The last one is a bit wordy, but you had no trouble understanding it. It may be clumsy style, but it's not ungrammatical.

(By the way, the example with sitting next to violates the artificial rule, imported from Latin, that prepositions should never come at the ends of sentences — but, again, it's perfectly natural English as the language is actually spoken. We're analyzing real English here, not the language found in grammar books.)

Now for the Coordinate Structure Constraint. Ross discovered that when the questioned noun phrase is one of a pair joined by and, the process that we've been describing does not take place.

You saw a duck and a goose.

What did you see a duck and _ ?    (ungrammatical)

What did you see _ and a goose?    (ungrammatical)

Not only is the Coordinate Structure Constraint quite strict, it appears to apply to all languages (at least, all languages in which we can identify a question-forming process like ours), not just English.

It seems to be more than just a rule of grammar — not something children acquire as they're learning to speak English, but rather, something deeper having to do with the mechanisms by which sentences are constructed.

A long line of research ensued from that discovery. But now Kubota and Lee have a completely different proposal.

They argue, based on evidence from Korean and Japanese, that the Coordinate Structure Constraint is not a rule or principle of grammar at all.

Instead, it is a rule of discourse structure: phrases joined by and have to be parallel. That is, they have to play the same role in communication, as alternatives or a pair.

Old-fashioned English teachers will be nodding and saying, "Parallelism!" Yes; that's exactly it. The Coordinate Structure Constraint is not a rule of grammar; it is a rule of how to express your ideas in an orderly way (the area of linguistics that is now called pragmatics).

Here's an example (based on Kubota and Lee) showing that parallel communicative intent is what matters. We sometimes use and to indicate logical succession rather than pairing. When we do, it's OK to question the second noun phrase:

I went to the store and bought milk.

What did you go to the store and buy _ ?

Here there is no violation of parallelism because go to the store and buy milk were not a matched pair in the first place; one of them was a consequence of the other.

Their evidence from Korean and Japanese is along the same lines but more extensive.

Now you know what linguists spend their time thinking about!

Permanent link to this entry