Malcolm Needs

Somewhere, in Patrick Mollison's work, cited in Blood Transfusion in Clinical Medicine, he mentions that IgG ABO antibodies are more clinically significant in solid organ transplants than are IgM (if I remember correctly, he specifically mentioned renal transplants), but I cannot cite the exact paper off the top of my head (I will see if I can find the reference).

As a result, whenever we were dealing with a renal transplant that crosses the ABO barrier, we almost performed an IgM and an a separate IgG titre.  Whether this is now considered to be necessary, I will leave to other people to discuss!

I did allo-adsorptions on eluates for quite a while and never once detected anything in the adsorbed eluate.  My own experience suggests that it is a waste of time and resources, but others may disagree.

I have been thinking about this and I have come, more or less, to the same way of thinking as Neil Blumberg.
The first pregnancy almost certainly could not have caused sensitisation of most of the common antigens, as some would not be formed on the foetal red cells at 10 weeks of gestation, while, even with a huge foeto-maternal haemorrhage (FMH) (in terms of the ratio of the total foetal red cell mass), the actual volume of foetal red cells transferred to the maternal circulation would not be sufficient to cause sensitisation in anyone but a person who is a "super responder".
The second pregnancy could easily have caused a primary response, either at partum, or as a result of a chronic FMH throughout much of the pregnancy.  Unless the woman's antibody screen was performed at about six months post-partum, such antibodies may never have been detected at their peak, and then may have declined to levels where normal serological techniques would not detect them.  Of course, all of this is theoretical, but, given the fact that the mother is probably an R1R1 (from information given above), it is most unlikely that an FMH estimation, such as a Kleihauer test was performed, or that the mother would have been serologically "followed up".  You also do not give the woman's transfusion history.
It would be helpful to know the time between the second and third pregnancy, and also, of course, if the same male was the biological father in both pregnancies.  This latter piece of information may actually be very difficult indeed to ascertain, as the woman may not wish to disclose her sexual history.

It is not unknown for antibodies of a particular specificity (say, purely for example, an anti-Jka) to increase in strength (as measured by serial titres) during a pregnancy that is, in this example, Jk(a-).  This is more common in a pregnancy where there is at least one other specificity (let us say, again, just as an example, anti-K), where (again, just as an example) the foetal red cells do express the K antigen, but not the Jk(a) antigen.
Lastly, antibodies that are forming de novo, very often seem to cross-react with antigens to which they are not actually stimulated.  This is true, even in the case of some monoclonal antibodies (particularly those within the HLA system), but some antibodies (again, even monoclonal antibodies) maintain what I will call a pseudo-specificity even in the "mature state".  This includes monoclonal anti-D.  Thorpe et al have reported that monoclonal anti-D molecules possess a V4-34 moiety, that is also present in anti-I and  anti-i, which is why these reagents should never be used straight from the fridge (Thorpe SJ, Boult CE, Stevenson FK, Scott ML, Sutherland J, Spellerberg MB, Natvig JB, Thompson KM.  Cold agglutinin activity is common among human monoclonal IgM Rh system antibodies using the V4-34 heavy chain variable gene segment.  Transfusion 1997; 37: 1111-1116 and Thorpe SJ, Ball C, Fox B, Thompson KM, Thorpe R, Bristow A.  Anti-D and anti-i activities are inseparable in V4-34-encoded monoclonal  anti-D: the same framework 1 residues are required for both activities.  Transfusion 2008; 48: 930-940).
I hope this helps a little.

It is all over the place, to be honest.

It is Caucasian, rather than caucasian,  It is group O, D Positive, and group A, D Positive, rather than either group O Positive or group A Positive (see the early editions of Peter Issitt's book).
It is Oh (with a subscript "h"), and not "Bombay".  The FUT1 gene, or, rather, the lack of a functional gene through various different genetic mutations, leads to the "Oh" phenotype, but this should NOT be called the "Bombay phenotype".  Although this phenotype was first described by Bhende YM, Deshpande CK, Bhatia HM, Sanger R, Race RR, Morgan WTJ, Watkins WM.  A “new” blood-group character related to the ABO system.  Lancet 1952; i: 903-904.  DOI:  10.1016/S0140-6736(52)92356-8, Another example of the Oh phenotype can be seen in the rare recessive condition, Leukocyte Adhesion Deficiency Type II where, to all intents and purposes, the patient will have a normal H gene, and yet the red cells are of the Oh phenotype, and anti-H can be found in the plasma. the phenotype has been identified in many different parts of the world (and is not just confined to mutations in India or even Asia (Hidalgo A, Ma S, Peired AJ, Weiss LA, Cunningham-Rundles C, Frenette PS.  Insights into leukocyte adhesion deficiency type 2 from a novel mutation in the GDP-fucose transporter gene.  Blood 2003; 101: 1705-1712.  DOI: 10.1182/blood-2002-09-2840).

The other thing is, of course, that "Bombay" no longer exists - it is now Mumbai!
I APOLOGISE FOR BEING A COMPLETE PEDANT!

The timing for fresh samples is somewhat different in the UK than in, for example, the USA.  The timings, and the reasons for these timings, are set out in paragraph 3.7 of the BCSH Guideline "Guidelines for pre-transfusion compatibility procedures in blood transfusion laboratories" (written by Claire Milkins, Jenny Berryman, Carol Cantwell, Chris Elliott, Richard Haggas, Joan Jones, Megan Rowley, Mark Williams and Nay Win, for, and on behalf of the BCSH, and published in Transfusion Medicine 2013; 23: 3-35.  doi: 10.1111/j.1365-3148.2012.01199.x), with the Key Recommendation of this paragraph being, "Serological studies should be performed using blood collected no more than 3 days in advance of the actual transfusion when the patient has been transfused or pregnant within the preceding 3 months."
As I understand it, this timing was based on the work originally carried out by Professor Patrick Mollison many years ago, who found that a new specificity of an alloantibody, or a nascent antibody that has become undetectable by normal serological techniques, can appear (or reappear) in the plasma of an individual within three days, after stimulation.  The problem is that not all patients know whether or not, or when, they have been transfused, or may deny it for religious reasons (I once cross-matched for a patient who had been a life-long Jehovah's Witness, who had an anti-Fya in his plasma that had a titre well in excess of 128).
I hope that helps.

The three months was chosen following a paper written by Laine EP, Leger RM, Arndt PA, Calhoun L, Garratty G, Petz LD. (In vitro studies of the impact of transfusion on the detection of alloantibodies after autoadsorption.  Transfusion 2000; 40 1384-1387.  DOI: 10.1046/j.1537-2995.2000.40111384.x.) that showed that red cells that had been transfused (or entered the circulation  via a feto-maternal haemorrhage could adsorb out weak alloantibodies for up to three months in a patient with AIHA.  This in vivo adsorption would, of course, also apply to individuals who did not have AIHA, but could lead to a secondary stimulation, leading to a stronger antibody (higher titre and higher concentration per mL of plasma), if the alloantibody was "missed" in the antibody screen and/or cross-match, particularly as it is unlikely that the full phenotype of the transfused (or foetal) red cells would be known.

Column Agglutination Technology (CAT) is superb at detecting IgG antibodies, and also at detecting certain "cold-reacting" antibodies (such as anti-I and (because of the pH of the column) anti-M, BUT the manufacturers themselves state quite freely, that they are not designed to detect all ABO mis-matches, because most ABO antibodies are IgM (anti-M are usually a mixture of IgM and IgG, but also react preferentially at a low pH).

In terms of the function of the various ABO blood types, there have been a huge number of peer-reviewed papers written on the subject (and the number has exploded with the advent of COVID19).  I would seriously defy anyone to keep up with all of these, but I would recommend reading pages 42-43 of Reid ME, Lomas-Francis C, Olsson ML.  The Blood Group Antigen FactsBook.  3rd edition, 2012. Academic Press.  ISBN: 978-0-12-415849-8.

In terms of how they evolved, it is so far back now that it is anyone's guess, but slides 28 to 32 of the attached lecture may give you some idea.
In Depth Lecture on The ABO and H Blood Group Systems.pptx

It simply means that the P1 antigen is particularly strongly expressed on these red cell samples.  Therefore, if you come across a weak anti-P1, it may apparently react with these particular red cell samples, whilst apparently not with, for example, the third red cell sample shown in your antigram.  Although not identical to dosage, per se, it is fairly synonymous with dosage at a phenotypical level.
The strength of the expression of the P1 antigen is an inherited trait.