intuition was never conclusive: the distances over which quarks roam might just get smaller and smaller, and the condensate less and less tenuous, with no abrupt jumps in the Uni-verse’s constitution. But it is hard to visualize appropriate intermediate states.

Fortunately, the equations of QCD are pre-cisely defined and reliable, so the question can be settled by calculation. Unfortunately, the equations are hard to solve, so until now a definitive answer has been elusive. By exploit-ing the full power of modern, large-scale com-puting, Aoki et al.1 have answered the question, ‘Did the Big Bang boil?’. The answer, as far as the quark–hadron transition is concerned, is ‘No’. QCD evolves smoothly with tem-perature; there is no thermodynamic phase transition.

Thus, Aoki and collaborators have brought closure to some intriguing speculations, and reinforced the foundations of Big Bang cosmology. For the powerful techniques they used, however, it is only the beginning.

According to a famous Berkeley graffito, “Reality is a crutch.” Although QCD with the values of quark masses that we have in our Universe apparently does not feature a true phase transition, it will be important to explore whether that conclusion remains true if those values are changed. There are quite firm and detailed theoretical predictions that phase transitions of both first and second order will occur in this case. (Second-order transi-tions are continuous, but feature more subtle singularities. They, too, can lead to large-scale density fluctuations.)

EVOLUTIONARY BIOLOGY

Fly eyes get the whole pictureKevin Moses

The compound eyes of ancestral flies picked up only one picture point in each facet. The evolution of a means to split up the light-sensitive cells increased this number to seven, boosting the eye’s resolution greatly.

“As anyone knows who has tried to catch one, flies see extremely well.” — Ian A. Meinertzhagen

Clearly, from the fly’s perspective, seeing well is a very good thing. Those ancestral flies that did not see well contributed more to the evolution of frogs than to the fly lineage. On page 696 of this issue, Zelhof and colleagues1 reveal how flies improved the resolution of their eyes dra-matically by evolving a specific modification to how their photoreceptor cells stick together*.

From fossils (mostly trilobites), we know that arthropods have used faceted, compound eyes ever since the Cambrian era, about 540

million years ago. Compound eyes produce a mosaic view of the environment because each of the eye facets (or ommatidia) is directed at a slightly different point in space. The resolu-tion of the image is limited by the number of ommatidia and the angles between them — to get more picture points requires more omma-tidia. Even though insect photoreceptor cells have evolved to be very small, there is a fitness cost to growing a bigger eye. So, for most of the past 540 million years, arthropods have had to deal with the problem that better vision demands a bigger eye.

But there are different vision requirements at night from those for the daytime, so the optimal balance in the trade-off between resolution and size varies. We humans solve this conundrum by

*This article and the paper concerned1 were published online on 1 October 2006.

Also, collisions of heavy ions at almost the speed of light are being studied at the Rela-tivistic Heavy Ion Collider at the Brookhaven National Laboratory on Long Island, and will soon be studied at CERN’s Large Hadron Col-lider near Geneva. Such collisions can produce fireballs with a significant excess of quarks over antiquarks, or different effective temperatures for quarks and gluons — possibilities that did not occur in the cosmic Big Bang. In those new circumstances, do true phase transitions occur? Once we learn the answers to such questions, a dramatic confrontation between theory and experiment might be staged.

Looking further ahead, we will progres-sively learn more about so-called electroweak symmetry breaking (the question of how the mediator particles of the electromagnetic and weak forces came to have different masses), the possible unification of all the fundamental forces, and other physics beyond the standard model. Then, the question of possible phase transitions other than that considered by Aoki et al.1, occurring at still higher temperatures, will come more sharply into focus. With luck, we will learn enough to compute definitive answers. In the meantime, we can have fun developing the necessary intellectual strength through exercise on QCD and its variants. ■ Frank Wilczek is in the Center for Theoretical Physics, Massachusetts Institute of Technology, Cambridge, Massachusetts 02142, USA.e-mail: wilczek@mit.edu

1. Aoki, Y., Endrodi, G., Fodor, Z., Katz, S. D. & Szabó, K. K. Nature 443, 675–678 (2006).

having two eyes built into each of ours. About 95% of our retina is covered in rod cells, which cannot distinguish colours but work well at night. For daylight vision, in the centre of our retina (the macula) about 5% of our photo-receptor cells (the cones) can distinguish colour but are less sensitive. Furthermore, our eyes are wired for low resolution at night, when it is more important to see every photon, and for high res-olution in daylight. We achieve this by having many rod cells stimulate one of the interneurons that process the retinal signal and pass it on to the brain, whereas in the centre of the retina each cone cell feeds only one interneuron.

Insects solve the day–night trade-off in a different way2. Insects that must see well in the daytime have ‘apposition’ eyes, with seven or eight photoreceptor cells clustered under each ommatidial lens. These cells are arranged so that their light-gathering surfaces (the rhab-domeres) are fused at the facet’s centre to make a structure called the rhabdom, so that they all receive light along the central optical axis (Fig. 1a). The entire facet is sheathed in a tube of pigment cells so that light coming in at other angles is removed (Fig. 1b). So most photons are lost, but resolution is improved. Night-flying insects have ‘superposition’ eyes, which lack most of the sheathing pigment so that light from a wider field can act on each photorecep-tor (Fig. 1c). This is more sensitive but sacri-fices resolution. Many insects are able to shift pigment within their pigment cells so that their eye can function in both ways: in effect, they raise their curtains at night.

About 100 million years ago, the diptera (mosquitoes and flies) developed a clever trick to squeeze more picture points out of their apposition eyes — they separated the photo-receptor rhabdomeres in each ommatidium so that, instead of seeing one picture point per facet, they now see seven (to the dismay of frogs ever since; Fig. 1d). This has been called a ‘neural superposition’ eye3 and also involves some equally clever rewiring in the fly’s brain4 (Fig. 1e). In their paper1, Zelhof and colleagues describe how diptera managed to develop this system.

Zelhof et al. looked for fruitfly mutants that have no inter-rhabdomeral space but have the rhabdomeres collapsed back into the centre of the facet like lower insects. They found loss-of-function mutations in two genes that have this effect, but otherwise leave the rhabdomeres intact. One gene they name spacemaker, and the other is a gene that encodes a previously known protein called Prominin that helps to form cell–cell contacts. The Spacemaker product was predicted to be a secreted protein, and Zelhof and colleagues show that it is indeed normally present in the inter-rhabdomeral space, begin-ning in the pupal stage when the rhabdomeres are formed. Furthermore, they show that Space-maker protein can bind to Prominin (in tissue-culture binding and cell-adhesion assays). Their data suggest that Spacemaker and Prominin normally function to oppose a third, previously

638

NATURE|Vol 443|12 October 2006NEWS & VIEWS

������������������� � ��������������������

Nature Publishing Group ©2006

Openrhabdom

Fusedrhabdomeres

a

Rhabdom

Lens

Pigment

Rhabdom

Lens

Pigment

b c

ed

50 &

100

YEA

RS A

GO

known protein, Chaoptin5, to divide the upper part of the photoreceptor cell membrane into rhabdomere and ‘stalk’ domains. These proteins also delicately regulate cell–cell contacts so that the rhabdoms stay intact, but the adjacent rhab-domeres don’t stick to each other.

Zelhof et al. find that Spacemaker is absent from the eyes of insects with fused rhabdom systems (bees and beetles), but is present in other diptera with open rhabdoms (the house fly and a mosquito). Prominin and Chaop-tin are present in all cases, and so it seems that Spacemaker is the crucial factor in mak-ing the inter-rhabdomeral space. In a most satisfying control experiment, they targeted the expression of Spacemaker to another type of fly eye: the ocelli (simple eyes on the top of the head). The three ocelli have many photo-receptor cells, but their rhabdomeres are nor-mally fused. When Zelhof et al. expressed

Spacemaker in the ocelli, they found that an inter-rhabdomeral space opens up.

These experiments suggest that the diptera may have ‘opened their eyes’ (invented neural superposition) by a single change: reprogram-ming the expression of Spacemaker for novel expression in the ommatidia. It is not often that we get such a clear glimpse of the blind watchmaker at work. ■

Kevin Moses is at the Howard Hughes Medical Institute, Janelia Farm Research Campus, 19700 Helix Drive, Ashburn, Virginia 20147, USA.e-mail: mosesk@hhmi.org

1. Zelhof, A. C., Hardy, R. W., Becker, A. & Zuker, C. S. Nature 443, 696–699 (2006).

2. Land, M. F. & Fernald, R. D. Annu. Rev. Neurosci. 15, 1–29 (1992).3. Franceschini, N. in Photoreceptor Optics (eds Snyder, A. W.

& Menzel, R.) 98–125 (Springer, Heidelberg, 1975).4. Meinertzhagen, I. A. Neuron 28, 310–313 (2000).5. van Vactor, D. L. J., Krantz, D. E., Reinke, R. & Zipursky, S. L.

Cell 52, 281–290 (1988).

50 YEARS AGOA new and rapid technique of characterizing the chemical properties of a protein in considerable detail has been devised; by its application a specific difference is found in the sequence of amino-acid residues of normal and sickle-cell haemoglobin. This difference appears to be confined to one small section of the polypeptide chains… The action of trypsin on proteinsis at present the most reliable way of splitting a peptide chainat specific peptide bonds…Small differences in the twoproteins will result in smallchanges in one or more of [the resulting] peptides. These should be detectable when the mixture is examined by a two-dimensional combination of paper electrophoresis and paper chromatography. It was decided to call the resulting chromatogram the ‘finger print’ of the protein. V. M. Ingram From Nature 13 October 1956.

100 YEARS AGOIn my address at York I urged biometricians to make sure that the problems they seek to elucidate are sound from a biological point of view. When asked by Prof. Pearson for an instance of failure in this respect I gave him, while away on my holiday, and in a private letter, Dr. Pearl’s paper. He has now seen fit, although I twice asked him to wait for a full answer until my return to Cambridge, to challenge me to show in the pages of Nature how my advice was applicable to that paper. I must leave your readers to judge how far I have succeeded in so doing.

The task has been far from an agreeable one. I should never have thought of singling Dr. Pearl’s paper out for public criticism in this manner had I not been challenged to do so. I can only say that if he feels himself aggrieved at the result, he can be in no doubt whom he has to thank. J. J. ListerFrom Nature 11 October 1906.

Figure 1 | How fly eyes differ from those of other arthropods. a, d, Surface view of a single eye facet (ommatidium) of a compound eye. b, c, e, Ommatidia seen as though sliced through their long axis. a, In the ancestral fly eye and in modern-day insects such as bees and beetles, the light-capturing surfaces (rhabdomeres) are fused in the centre of the ommatidium. b, In ‘apposition’ eyes, pigment cells sheath each ommatidium, so light is received only down the central axis and image resolution is improved. c, Insects that need good night vision have ‘superposition’ eyes, which lack most of this pigment, so light is received from all angles. In this case more photons are received, but the resolution of the image is reduced. d, Modern flies and some mosquitoes have open rhabdoms, so that instead of one picture point per ommatidium, they can perceive seven. This requires some rewiring of the underlying neurons so that light received from different angles in one ommatidium can be resolved (neural superposition, e). Zelhof et al.1 find that expression of the Spacemaker protein is responsible for the opening up of the rhabdomeres.

639

NATURE|Vol 443|12 October 2006 NEWS & VIEWS

������������������� � ��������������������

Nature Publishing Group ©2006