What happened. The gene entered the cells but was never linked to a replicating sequence. Most likely such a piece of DNA is not able to multiply itself in the progeny cells of the organism, because the cell's machinery replicates chromosomes, not loose fragments. As the culture divided, the original copies were distributed among an exponentially growing number of cells and then degraded, so by day 4 the gene was below detection.

Why the bacteria were unaffected. The gene was never expressed in a way that mattered to them, and its loss cost them nothing. Normal growth is exactly what you would expect.

The single change. Link the gene to an origin of replication before transfer — in practice, ligate it into a plasmid vector. An alien piece of DNA joined to an ori can replicate and multiply itself in the host organism. (Integration into the host genome would work too, since the DNA then becomes part of a chromosome that has the ability to replicate, but that is not something the student can arrange simply by pushing DNA into cells.)

What genetic engineering adds. Traditional hybridisation procedures used in plant and animal breeding very often lead to inclusion and multiplication of undesirable genes along with the desired genes, because entire genomes are recombined and the breeder has no control over which segments travel together. Genetic engineering allows us to isolate and introduce only one or a set of desirable genes without introducing undesirable genes into the target organism. It also ignores the species barrier: a gene can be moved from a human to a bacterium, which no amount of crossing could achieve.

What hybridisation still does better. It improves traits controlled by many genes at once — yield, hardiness, flavour, tolerance of poor soil — which no one can specify as a list of genes to transfer. It also tests the whole organism in a real field every generation, so unfit combinations are eliminated automatically. Genetic engineering is precise about a small target; breeding is blunt but works on the whole plant.

The honest conclusion. They are complementary. A gene of interest is usually introduced by engineering and then moved into locally adapted varieties by conventional crossing.

There is a real principle. The EFB definition is 'the integration of natural science and organisms, cells, parts thereof, and molecular analogues for products and services'. Both cases satisfy it: a living organism, or a part of one, is deliberately used to produce something people want. Making curd, bread or wine are all microbe-mediated processes and could also be thought of as a form of biotechnology.

Where they differ. The restricted modern sense of the term refers to processes which use genetically modified organisms to achieve the same on a larger scale. Curd uses an organism as found; recombinant insulin uses an organism rebuilt for the purpose, in a bioreactor, under bioprocess engineering control.

So the answer is both. The principle is common — harness a living system for a product. The distinction worth making is not biotechnology versus not-biotechnology, but traditional versus modern molecular biotechnology, and the EFB definition was framed wide precisely so that it encompasses both views.

Bioprocess engineering is failing, not genetic engineering. The construct works — the flask proves it. What is not being maintained is the sterile, microbial contamination-free ambience that enables growth of only the desired microbe in large quantities.

What is likely going wrong. At 500 litres, a contaminant that would be trivial in a flask has two days of exponential growth in a nutrient-rich vessel; it outgrows the engineered strain, which is usually the slower grower because it carries an extra plasmid and an extra protein to make. Alongside contamination, the large vessel must also supply what the flask supplied passively — even mixing, oxygen throughout the volume, and control of temperature and pH — and a failure of any of these stops production even with no contaminant at all.

The general point. Modern biotechnology needed two inventions, not one. Genetic engineering gives you a cell that can make the product; bioprocess engineering is what turns that cell into a tonne of product.

Because its success is visible without any further test. The experiment had to answer one question: did the alien DNA get in, and is it being maintained and copied? An antibiotic resistance gene answers that question by itself. Spread the cells on a medium containing the antibiotic; a colony that grows must be carrying a working copy of the gene, and one that does not carry it dies. No sequencing, no protein assay, nothing.

Three further reasons. First, it was available and already localised — the resistance gene could be cut out of a plasmid known to confer resistance. Second, it works in the host used: normal E. coli cells do not carry resistance to these antibiotics, so there is no background of false positives. Third, the readout is quantitative — the number of resistant colonies measures how well the whole procedure worked.

And it set a lasting pattern. The gene that proved the principle in 1972 is still in every vector you will meet in the next part, now under a different name: the selectable marker.